diff options
author | Sebastian Andrzej Siewior <bigeasy@linutronix.de> | 2021-09-07 12:43:49 +0200 |
---|---|---|
committer | Sebastian Andrzej Siewior <bigeasy@linutronix.de> | 2021-09-07 12:43:49 +0200 |
commit | 5beb0b705c1e1173ebf6e53d33f7f5e0045a2835 (patch) | |
tree | 4d4c2c36ae35ed6445fab729cb266cf6decb416e | |
parent | 9985ec69e335f755bcb2e583344155fa6ef914bd (diff) | |
download | linux-rt-8a35f23f0c85f67f2d26bf78aa138d7dcd0bda60.tar.gz |
[ANNOUNCE] v5.14.1-rt19v5.14.1-rt19-patches
Dear RT folks!
I'm pleased to announce the v5.14.1-rt19 patch set.
Changes since v5.14.1-rt18:
- Dan Carpenter reported a possible NULL pointer dereference in
rt_mutex_adjust_prio_chain(). Patch by Peter Zijlstra.
- Unused rt_rwlock_is_contended() has been removed. Reported by kernel
test robot.
- The "play idle" timer was missing a _HARD annotation. It would
freeze the system uppon activation of the intel powerclamp driver.
Reported by Thomas Gleixner.
- Vlastimil Babka SLUB queue has been updated to v6r2.
Known issues
- netconsole triggers WARN.
- The "Memory controller" (CONFIG_MEMCG) has been disabled.
- A RCU and ARM64 warning has been fixed by Valentin Schneider. It is
still not clear if the RCU related change is correct.
- Clark Williams reported issues in i915 (execlists_dequeue_irq())
- Valentin Schneider reported a few splats on ARM64, see
https://https://lkml.kernel.org/r/.kernel.org/lkml/20210810134127.1394269-1-valentin.schneider@arm.com/
The delta patch against v5.14.1-rt18 is appended below and can be found here:
https://cdn.kernel.org/pub/linux/kernel/projects/rt/5.14/incr/patch-5.14.1-rt18-rt19.patch.xz
You can get this release via the git tree at:
git://git.kernel.org/pub/scm/linux/kernel/git/rt/linux-rt-devel.git v5.14.1-rt19
The RT patch against v5.14.1 can be found here:
https://cdn.kernel.org/pub/linux/kernel/projects/rt/5.14/older/patch-5.14.1-rt19.patch.xz
The split quilt queue is available at:
https://cdn.kernel.org/pub/linux/kernel/projects/rt/5.14/older/patches-5.14.1-rt19.tar.xz
Sebastian
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
42 files changed, 623 insertions, 663 deletions
diff --git a/patches/0001-mm-slub-don-t-call-flush_all-from-slab_debug_trace_o.patch b/patches/0001-mm-slub-don-t-call-flush_all-from-slab_debug_trace_o.patch index 1e0a3c0e60c0..0c84435c45a3 100644 --- a/patches/0001-mm-slub-don-t-call-flush_all-from-slab_debug_trace_o.patch +++ b/patches/0001-mm-slub-don-t-call-flush_all-from-slab_debug_trace_o.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:20:58 +0200 -Subject: [PATCH 01/35] mm, slub: don't call flush_all() from +Date: Fri, 28 May 2021 14:32:10 +0200 +Subject: [PATCH 01/33] mm, slub: don't call flush_all() from slab_debug_trace_open() slab_debug_trace_open() can only be called on caches with SLAB_STORE_USER flag @@ -9,7 +9,6 @@ slabs altogether, so there's nothing to flush. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 3 --- 1 file changed, 3 deletions(-) diff --git a/patches/0002-mm-slub-allocate-private-object-map-for-debugfs-list.patch b/patches/0002-mm-slub-allocate-private-object-map-for-debugfs-list.patch index 1d9cf5525b15..4900e14e9304 100644 --- a/patches/0002-mm-slub-allocate-private-object-map-for-debugfs-list.patch +++ b/patches/0002-mm-slub-allocate-private-object-map-for-debugfs-list.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:20:59 +0200 -Subject: [PATCH 02/35] mm, slub: allocate private object map for debugfs +Date: Sun, 23 May 2021 01:28:37 +0200 +Subject: [PATCH 02/33] mm, slub: allocate private object map for debugfs listings Slub has a static spinlock protected bitmap for marking which objects are on @@ -16,7 +16,6 @@ to use a private bitmap. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 44 +++++++++++++++++++++++++++++--------------- 1 file changed, 29 insertions(+), 15 deletions(-) diff --git a/patches/0003-mm-slub-allocate-private-object-map-for-validate_sla.patch b/patches/0003-mm-slub-allocate-private-object-map-for-validate_sla.patch index dd9c26775a55..6fc0b18f9099 100644 --- a/patches/0003-mm-slub-allocate-private-object-map-for-validate_sla.patch +++ b/patches/0003-mm-slub-allocate-private-object-map-for-validate_sla.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:00 +0200 -Subject: [PATCH 03/35] mm, slub: allocate private object map for +Date: Sun, 23 May 2021 01:37:07 +0200 +Subject: [PATCH 03/33] mm, slub: allocate private object map for validate_slab_cache() validate_slab_cache() is called either to handle a sysfs write, or from a @@ -11,7 +11,6 @@ critical sections, so let's do that. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 24 +++++++++++++++--------- 1 file changed, 15 insertions(+), 9 deletions(-) diff --git a/patches/0004-mm-slub-don-t-disable-irq-for-debug_check_no_locks_f.patch b/patches/0004-mm-slub-don-t-disable-irq-for-debug_check_no_locks_f.patch index aebe358c82ba..f2c670fa2b35 100644 --- a/patches/0004-mm-slub-don-t-disable-irq-for-debug_check_no_locks_f.patch +++ b/patches/0004-mm-slub-don-t-disable-irq-for-debug_check_no_locks_f.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:01 +0200 -Subject: [PATCH 04/35] mm, slub: don't disable irq for +Date: Fri, 21 May 2021 01:25:06 +0200 +Subject: [PATCH 04/33] mm, slub: don't disable irq for debug_check_no_locks_freed() In slab_free_hook() we disable irqs around the debug_check_no_locks_freed() @@ -13,7 +13,6 @@ Mel noted: Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 14 +------------- 1 file changed, 1 insertion(+), 13 deletions(-) diff --git a/patches/0005-mm-slub-remove-redundant-unfreeze_partials-from-put_.patch b/patches/0005-mm-slub-remove-redundant-unfreeze_partials-from-put_.patch index e9adac1c5455..a07027ba2f30 100644 --- a/patches/0005-mm-slub-remove-redundant-unfreeze_partials-from-put_.patch +++ b/patches/0005-mm-slub-remove-redundant-unfreeze_partials-from-put_.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:02 +0200 -Subject: [PATCH 05/35] mm, slub: remove redundant unfreeze_partials() from +Date: Tue, 8 Jun 2021 01:19:03 +0200 +Subject: [PATCH 05/33] mm, slub: remove redundant unfreeze_partials() from put_cpu_partial() Commit d6e0b7fa1186 ("slub: make dead caches discard free slabs immediately") @@ -17,7 +17,6 @@ unfreeze_partials() which could be thus also considered unnecessary. But further patches will rely on it, so keep it. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 7 ------- 1 file changed, 7 deletions(-) diff --git a/patches/0007-mm-slub-extract-get_partial-from-new_slab_objects.patch b/patches/0006-mm-slub-extract-get_partial-from-new_slab_objects.patch index 38665c4767ff..6da43addac38 100644 --- a/patches/0007-mm-slub-extract-get_partial-from-new_slab_objects.patch +++ b/patches/0006-mm-slub-extract-get_partial-from-new_slab_objects.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:04 +0200 -Subject: [PATCH 07/35] mm, slub: extract get_partial() from new_slab_objects() +Date: Tue, 11 May 2021 12:45:48 +0200 +Subject: [PATCH 06/33] mm, slub: extract get_partial() from new_slab_objects() The later patches will need more fine grained control over individual actions in ___slab_alloc(), the only caller of new_slab_objects(), so this is a first @@ -11,14 +11,13 @@ useful for later changes. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2599,17 +2599,12 @@ slab_out_of_memory(struct kmem_cache *s, +@@ -2613,17 +2613,12 @@ slab_out_of_memory(struct kmem_cache *s, static inline void *new_slab_objects(struct kmem_cache *s, gfp_t flags, int node, struct kmem_cache_cpu **pc) { @@ -37,7 +36,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> page = new_slab(s, flags, node); if (page) { c = raw_cpu_ptr(s->cpu_slab); -@@ -2773,6 +2768,10 @@ static void *___slab_alloc(struct kmem_c +@@ -2787,6 +2782,10 @@ static void *___slab_alloc(struct kmem_c goto redo; } @@ -48,7 +47,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> freelist = new_slab_objects(s, gfpflags, node, &c); if (unlikely(!freelist)) { -@@ -2780,6 +2779,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2794,6 +2793,7 @@ static void *___slab_alloc(struct kmem_c return NULL; } diff --git a/patches/0006-mm-slub-unify-cmpxchg_double_slab-and-__cmpxchg_doub.patch b/patches/0006-mm-slub-unify-cmpxchg_double_slab-and-__cmpxchg_doub.patch deleted file mode 100644 index b76ba63b77ba..000000000000 --- a/patches/0006-mm-slub-unify-cmpxchg_double_slab-and-__cmpxchg_doub.patch +++ /dev/null @@ -1,122 +0,0 @@ -From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:03 +0200 -Subject: [PATCH 06/35] mm, slub: unify cmpxchg_double_slab() and - __cmpxchg_double_slab() - -These functions differ only in irq disabling in the slow path. We can create a -common function with an extra bool parameter to control the irq disabling. -As the functions are inline and the parameter compile-time constant, there -will be no runtime overhead due to this change. - -Also change the DEBUG_VM based irqs disable assert to the more standard -lockdep_assert based one. - -Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Acked-by: Christoph Lameter <cl@linux.com> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> ---- - mm/slub.c | 62 ++++++++++++++++++++++++-------------------------------------- - 1 file changed, 24 insertions(+), 38 deletions(-) - ---- a/mm/slub.c -+++ b/mm/slub.c -@@ -371,13 +371,13 @@ static __always_inline void slab_unlock( - __bit_spin_unlock(PG_locked, &page->flags); - } - --/* Interrupts must be disabled (for the fallback code to work right) */ --static inline bool __cmpxchg_double_slab(struct kmem_cache *s, struct page *page, -+static inline bool ___cmpxchg_double_slab(struct kmem_cache *s, struct page *page, - void *freelist_old, unsigned long counters_old, - void *freelist_new, unsigned long counters_new, -- const char *n) -+ const char *n, bool disable_irqs) - { -- VM_BUG_ON(!irqs_disabled()); -+ if (!disable_irqs) -+ lockdep_assert_irqs_disabled(); - #if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \ - defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE) - if (s->flags & __CMPXCHG_DOUBLE) { -@@ -388,15 +388,23 @@ static inline bool __cmpxchg_double_slab - } else - #endif - { -+ unsigned long flags; -+ -+ if (disable_irqs) -+ local_irq_save(flags); - slab_lock(page); - if (page->freelist == freelist_old && - page->counters == counters_old) { - page->freelist = freelist_new; - page->counters = counters_new; - slab_unlock(page); -+ if (disable_irqs) -+ local_irq_restore(flags); - return true; - } - slab_unlock(page); -+ if (disable_irqs) -+ local_irq_restore(flags); - } - - cpu_relax(); -@@ -409,45 +417,23 @@ static inline bool __cmpxchg_double_slab - return false; - } - --static inline bool cmpxchg_double_slab(struct kmem_cache *s, struct page *page, -+/* Interrupts must be disabled (for the fallback code to work right) */ -+static inline bool __cmpxchg_double_slab(struct kmem_cache *s, struct page *page, - void *freelist_old, unsigned long counters_old, - void *freelist_new, unsigned long counters_new, - const char *n) - { --#if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \ -- defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE) -- if (s->flags & __CMPXCHG_DOUBLE) { -- if (cmpxchg_double(&page->freelist, &page->counters, -- freelist_old, counters_old, -- freelist_new, counters_new)) -- return true; -- } else --#endif -- { -- unsigned long flags; -- -- local_irq_save(flags); -- slab_lock(page); -- if (page->freelist == freelist_old && -- page->counters == counters_old) { -- page->freelist = freelist_new; -- page->counters = counters_new; -- slab_unlock(page); -- local_irq_restore(flags); -- return true; -- } -- slab_unlock(page); -- local_irq_restore(flags); -- } -- -- cpu_relax(); -- stat(s, CMPXCHG_DOUBLE_FAIL); -- --#ifdef SLUB_DEBUG_CMPXCHG -- pr_info("%s %s: cmpxchg double redo ", n, s->name); --#endif -+ return ___cmpxchg_double_slab(s, page, freelist_old, counters_old, -+ freelist_new, counters_new, n, false); -+} - -- return false; -+static inline bool cmpxchg_double_slab(struct kmem_cache *s, struct page *page, -+ void *freelist_old, unsigned long counters_old, -+ void *freelist_new, unsigned long counters_new, -+ const char *n) -+{ -+ return ___cmpxchg_double_slab(s, page, freelist_old, counters_old, -+ freelist_new, counters_new, n, true); - } - - #ifdef CONFIG_SLUB_DEBUG diff --git a/patches/0008-mm-slub-dissolve-new_slab_objects-into-___slab_alloc.patch b/patches/0007-mm-slub-dissolve-new_slab_objects-into-___slab_alloc.patch index f9434d20cd3e..9deaec0605cc 100644 --- a/patches/0008-mm-slub-dissolve-new_slab_objects-into-___slab_alloc.patch +++ b/patches/0007-mm-slub-dissolve-new_slab_objects-into-___slab_alloc.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:05 +0200 -Subject: [PATCH 08/35] mm, slub: dissolve new_slab_objects() into +Date: Tue, 11 May 2021 13:01:34 +0200 +Subject: [PATCH 07/33] mm, slub: dissolve new_slab_objects() into ___slab_alloc() The later patches will need more fine grained control over individual actions @@ -14,14 +14,13 @@ able to catch a development change introducing a systematic misuse. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Christoph Lameter <cl@linux.com> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 50 ++++++++++++++++++-------------------------------- 1 file changed, 18 insertions(+), 32 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -1871,6 +1871,8 @@ static struct page *new_slab(struct kmem +@@ -1885,6 +1885,8 @@ static struct page *new_slab(struct kmem if (unlikely(flags & GFP_SLAB_BUG_MASK)) flags = kmalloc_fix_flags(flags); @@ -30,7 +29,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> return allocate_slab(s, flags & (GFP_RECLAIM_MASK | GFP_CONSTRAINT_MASK), node); } -@@ -2596,36 +2598,6 @@ slab_out_of_memory(struct kmem_cache *s, +@@ -2610,36 +2612,6 @@ slab_out_of_memory(struct kmem_cache *s, #endif } @@ -67,7 +66,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> static inline bool pfmemalloc_match(struct page *page, gfp_t gfpflags) { if (unlikely(PageSlabPfmemalloc(page))) -@@ -2772,13 +2744,27 @@ static void *___slab_alloc(struct kmem_c +@@ -2786,13 +2758,27 @@ static void *___slab_alloc(struct kmem_c if (freelist) goto check_new_page; diff --git a/patches/0009-mm-slub-return-slab-page-from-get_partial-and-set-c-.patch b/patches/0008-mm-slub-return-slab-page-from-get_partial-and-set-c-.patch index 2c06299abe75..eb941ec46be4 100644 --- a/patches/0009-mm-slub-return-slab-page-from-get_partial-and-set-c-.patch +++ b/patches/0008-mm-slub-return-slab-page-from-get_partial-and-set-c-.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:06 +0200 -Subject: [PATCH 09/35] mm, slub: return slab page from get_partial() and set +Date: Tue, 11 May 2021 14:05:22 +0200 +Subject: [PATCH 08/33] mm, slub: return slab page from get_partial() and set c->page afterwards The function get_partial() finds a suitable page on a partial list, acquires @@ -12,14 +12,13 @@ kmem_cache_cpu.page pointer. No functional change as all of this still happens with disabled IRQs. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 21 +++++++++++---------- 1 file changed, 11 insertions(+), 10 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2003,7 +2003,7 @@ static inline bool pfmemalloc_match(stru +@@ -2017,7 +2017,7 @@ static inline bool pfmemalloc_match(stru * Try to allocate a partial slab from a specific node. */ static void *get_partial_node(struct kmem_cache *s, struct kmem_cache_node *n, @@ -28,7 +27,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> { struct page *page, *page2; void *object = NULL; -@@ -2032,7 +2032,7 @@ static void *get_partial_node(struct kme +@@ -2046,7 +2046,7 @@ static void *get_partial_node(struct kme available += objects; if (!object) { @@ -37,7 +36,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> stat(s, ALLOC_FROM_PARTIAL); object = t; } else { -@@ -2052,7 +2052,7 @@ static void *get_partial_node(struct kme +@@ -2066,7 +2066,7 @@ static void *get_partial_node(struct kme * Get a page from somewhere. Search in increasing NUMA distances. */ static void *get_any_partial(struct kmem_cache *s, gfp_t flags, @@ -46,7 +45,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> { #ifdef CONFIG_NUMA struct zonelist *zonelist; -@@ -2094,7 +2094,7 @@ static void *get_any_partial(struct kmem +@@ -2108,7 +2108,7 @@ static void *get_any_partial(struct kmem if (n && cpuset_zone_allowed(zone, flags) && n->nr_partial > s->min_partial) { @@ -55,7 +54,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (object) { /* * Don't check read_mems_allowed_retry() -@@ -2116,7 +2116,7 @@ static void *get_any_partial(struct kmem +@@ -2130,7 +2130,7 @@ static void *get_any_partial(struct kmem * Get a partial page, lock it and return it. */ static void *get_partial(struct kmem_cache *s, gfp_t flags, int node, @@ -64,7 +63,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> { void *object; int searchnode = node; -@@ -2124,11 +2124,11 @@ static void *get_partial(struct kmem_cac +@@ -2138,11 +2138,11 @@ static void *get_partial(struct kmem_cac if (node == NUMA_NO_NODE) searchnode = numa_mem_id(); @@ -78,7 +77,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } #ifdef CONFIG_PREEMPTION -@@ -2740,9 +2740,11 @@ static void *___slab_alloc(struct kmem_c +@@ -2754,9 +2754,11 @@ static void *___slab_alloc(struct kmem_c goto redo; } @@ -92,7 +91,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> page = new_slab(s, gfpflags, node); -@@ -2766,7 +2768,6 @@ static void *___slab_alloc(struct kmem_c +@@ -2780,7 +2782,6 @@ static void *___slab_alloc(struct kmem_c c->page = page; check_new_page: diff --git a/patches/0010-mm-slub-restructure-new-page-checks-in-___slab_alloc.patch b/patches/0009-mm-slub-restructure-new-page-checks-in-___slab_alloc.patch index b5511a0f418e..00b71cf553b5 100644 --- a/patches/0010-mm-slub-restructure-new-page-checks-in-___slab_alloc.patch +++ b/patches/0009-mm-slub-restructure-new-page-checks-in-___slab_alloc.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:07 +0200 -Subject: [PATCH 10/35] mm, slub: restructure new page checks in +Date: Tue, 11 May 2021 18:25:09 +0200 +Subject: [PATCH 09/33] mm, slub: restructure new page checks in ___slab_alloc() When we allocate slab object from a newly acquired page (from node's partial @@ -14,14 +14,13 @@ No functional change. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 28 ++++++++++++++++++++++------ 1 file changed, 22 insertions(+), 6 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2768,13 +2768,29 @@ static void *___slab_alloc(struct kmem_c +@@ -2782,13 +2782,29 @@ static void *___slab_alloc(struct kmem_c c->page = page; check_new_page: diff --git a/patches/0011-mm-slub-simplify-kmem_cache_cpu-and-tid-setup.patch b/patches/0010-mm-slub-simplify-kmem_cache_cpu-and-tid-setup.patch index 37f53c3424ae..8bb5b3f0f758 100644 --- a/patches/0011-mm-slub-simplify-kmem_cache_cpu-and-tid-setup.patch +++ b/patches/0010-mm-slub-simplify-kmem_cache_cpu-and-tid-setup.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:08 +0200 -Subject: [PATCH 11/35] mm, slub: simplify kmem_cache_cpu and tid setup +Date: Tue, 18 May 2021 02:01:39 +0200 +Subject: [PATCH 10/33] mm, slub: simplify kmem_cache_cpu and tid setup In slab_alloc_node() and do_slab_free() fastpaths we need to guarantee that our kmem_cache_cpu pointer is from the same cpu as the tid value. Currently @@ -16,14 +16,13 @@ freelist didn't change by anyone preempting us since reading the tid. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 22 +++++++++------------- 1 file changed, 9 insertions(+), 13 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2868,15 +2868,14 @@ static __always_inline void *slab_alloc_ +@@ -2882,15 +2882,14 @@ static __always_inline void *slab_alloc_ * reading from one cpu area. That does not matter as long * as we end up on the original cpu again when doing the cmpxchg. * @@ -46,7 +45,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * Irqless object alloc/free algorithm used here depends on sequence -@@ -3150,11 +3149,8 @@ static __always_inline void do_slab_free +@@ -3164,11 +3163,8 @@ static __always_inline void do_slab_free * data is retrieved via this pointer. If we are on the same cpu * during the cmpxchg then the free will succeed. */ diff --git a/patches/0012-mm-slub-move-disabling-enabling-irqs-to-___slab_allo.patch b/patches/0011-mm-slub-move-disabling-enabling-irqs-to-___slab_allo.patch index 73ca5d4db92c..a66be68ec007 100644 --- a/patches/0012-mm-slub-move-disabling-enabling-irqs-to-___slab_allo.patch +++ b/patches/0011-mm-slub-move-disabling-enabling-irqs-to-___slab_allo.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:09 +0200 -Subject: [PATCH 12/35] mm, slub: move disabling/enabling irqs to +Date: Fri, 7 May 2021 19:32:31 +0200 +Subject: [PATCH 11/33] mm, slub: move disabling/enabling irqs to ___slab_alloc() Currently __slab_alloc() disables irqs around the whole ___slab_alloc(). This @@ -37,14 +37,13 @@ the re-read of kmem_cache_cpu pointer is still compiled out as it was before. [ Mike Galbraith <efault@gmx.de>: Fix kmem_cache_alloc_bulk() error path ] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 36 ++++++++++++++++++++++++------------ 1 file changed, 24 insertions(+), 12 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2656,7 +2656,7 @@ static inline void *get_freelist(struct +@@ -2670,7 +2670,7 @@ static inline void *get_freelist(struct * we need to allocate a new slab. This is the slowest path since it involves * a call to the page allocator and the setup of a new slab. * @@ -53,7 +52,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> * already disabled (which is the case for bulk allocation). */ static void *___slab_alloc(struct kmem_cache *s, gfp_t gfpflags, int node, -@@ -2664,9 +2664,11 @@ static void *___slab_alloc(struct kmem_c +@@ -2678,9 +2678,11 @@ static void *___slab_alloc(struct kmem_c { void *freelist; struct page *page; @@ -65,7 +64,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> page = c->page; if (!page) { /* -@@ -2729,6 +2731,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2743,6 +2745,7 @@ static void *___slab_alloc(struct kmem_c VM_BUG_ON(!c->page->frozen); c->freelist = get_freepointer(s, freelist); c->tid = next_tid(c->tid); @@ -73,7 +72,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> return freelist; new_slab: -@@ -2746,14 +2749,16 @@ static void *___slab_alloc(struct kmem_c +@@ -2760,14 +2763,16 @@ static void *___slab_alloc(struct kmem_c goto check_new_page; } @@ -91,7 +90,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (c->page) flush_slab(s, c); -@@ -2793,31 +2798,33 @@ static void *___slab_alloc(struct kmem_c +@@ -2807,31 +2812,33 @@ static void *___slab_alloc(struct kmem_c return_single: deactivate_slab(s, page, get_freepointer(s, freelist), c); @@ -133,7 +132,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> return p; } -@@ -3345,8 +3352,8 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3359,8 +3366,8 @@ int kmem_cache_alloc_bulk(struct kmem_ca * IRQs, which protects against PREEMPT and interrupts * handlers invoking normal fastpath. */ @@ -143,7 +142,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> for (i = 0; i < size; i++) { void *object = kfence_alloc(s, s->object_size, flags); -@@ -3367,6 +3374,8 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3381,6 +3388,8 @@ int kmem_cache_alloc_bulk(struct kmem_ca */ c->tid = next_tid(c->tid); @@ -152,7 +151,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * Invoking slow path likely have side-effect * of re-populating per CPU c->freelist -@@ -3379,6 +3388,8 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3393,6 +3402,8 @@ int kmem_cache_alloc_bulk(struct kmem_ca c = this_cpu_ptr(s->cpu_slab); maybe_wipe_obj_freeptr(s, p[i]); @@ -161,7 +160,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> continue; /* goto for-loop */ } c->freelist = get_freepointer(s, object); -@@ -3387,6 +3398,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3401,6 +3412,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca } c->tid = next_tid(c->tid); local_irq_enable(); @@ -169,7 +168,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * memcg and kmem_cache debug support and memory initialization. -@@ -3396,7 +3408,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3410,7 +3422,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca slab_want_init_on_alloc(flags, s)); return i; error: diff --git a/patches/0013-mm-slub-do-initial-checks-in-___slab_alloc-with-irqs.patch b/patches/0012-mm-slub-do-initial-checks-in-___slab_alloc-with-irqs.patch index d9c1bc411bab..2e9b8e7fd07d 100644 --- a/patches/0013-mm-slub-do-initial-checks-in-___slab_alloc-with-irqs.patch +++ b/patches/0012-mm-slub-do-initial-checks-in-___slab_alloc-with-irqs.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:10 +0200 -Subject: [PATCH 13/35] mm, slub: do initial checks in ___slab_alloc() with +Date: Sat, 8 May 2021 02:28:02 +0200 +Subject: [PATCH 12/33] mm, slub: do initial checks in ___slab_alloc() with irqs enabled As another step of shortening irq disabled sections in ___slab_alloc(), delay @@ -17,7 +17,6 @@ variant that lacks the PageSlab check. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Acked-by: Mel Gorman <mgorman@techsingularity.net> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- include/linux/page-flags.h | 9 +++++++ mm/slub.c | 54 +++++++++++++++++++++++++++++++++++++-------- @@ -43,7 +42,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> VM_BUG_ON_PAGE(!PageSlab(page), page); --- a/mm/slub.c +++ b/mm/slub.c -@@ -2607,6 +2607,19 @@ static inline bool pfmemalloc_match(stru +@@ -2621,6 +2621,19 @@ static inline bool pfmemalloc_match(stru } /* @@ -63,7 +62,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> * Check the page->freelist of a page and either transfer the freelist to the * per cpu freelist or deactivate the page. * -@@ -2668,8 +2681,9 @@ static void *___slab_alloc(struct kmem_c +@@ -2682,8 +2695,9 @@ static void *___slab_alloc(struct kmem_c stat(s, ALLOC_SLOWPATH); @@ -75,7 +74,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (!page) { /* * if the node is not online or has no normal memory, just -@@ -2678,6 +2692,11 @@ static void *___slab_alloc(struct kmem_c +@@ -2692,6 +2706,11 @@ static void *___slab_alloc(struct kmem_c if (unlikely(node != NUMA_NO_NODE && !node_isset(node, slab_nodes))) node = NUMA_NO_NODE; @@ -87,7 +86,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> goto new_slab; } redo: -@@ -2692,8 +2711,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2706,8 +2725,7 @@ static void *___slab_alloc(struct kmem_c goto redo; } else { stat(s, ALLOC_NODE_MISMATCH); @@ -97,7 +96,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } } -@@ -2702,12 +2720,15 @@ static void *___slab_alloc(struct kmem_c +@@ -2716,12 +2734,15 @@ static void *___slab_alloc(struct kmem_c * PFMEMALLOC but right now, we are losing the pfmemalloc * information when the page leaves the per-cpu allocator */ @@ -118,7 +117,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> freelist = c->freelist; if (freelist) goto load_freelist; -@@ -2723,6 +2744,9 @@ static void *___slab_alloc(struct kmem_c +@@ -2737,6 +2758,9 @@ static void *___slab_alloc(struct kmem_c stat(s, ALLOC_REFILL); load_freelist: @@ -128,7 +127,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * freelist is pointing to the list of objects to be used. * page is pointing to the page from which the objects are obtained. -@@ -2734,11 +2758,23 @@ static void *___slab_alloc(struct kmem_c +@@ -2748,11 +2772,23 @@ static void *___slab_alloc(struct kmem_c local_irq_restore(flags); return freelist; diff --git a/patches/0014-mm-slub-move-disabling-irqs-closer-to-get_partial-in.patch b/patches/0013-mm-slub-move-disabling-irqs-closer-to-get_partial-in.patch index a4baed5461da..ef3f9b15596c 100644 --- a/patches/0014-mm-slub-move-disabling-irqs-closer-to-get_partial-in.patch +++ b/patches/0013-mm-slub-move-disabling-irqs-closer-to-get_partial-in.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:11 +0200 -Subject: [PATCH 14/35] mm, slub: move disabling irqs closer to get_partial() +Date: Mon, 10 May 2021 13:56:17 +0200 +Subject: [PATCH 13/33] mm, slub: move disabling irqs closer to get_partial() in ___slab_alloc() Continue reducing the irq disabled scope. Check for per-cpu partial slabs with @@ -8,14 +8,13 @@ first with irqs enabled and then recheck with irqs disabled before grabbing the slab page. Mostly preparatory for the following patches. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 34 +++++++++++++++++++++++++--------- 1 file changed, 25 insertions(+), 9 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2692,11 +2692,6 @@ static void *___slab_alloc(struct kmem_c +@@ -2706,11 +2706,6 @@ static void *___slab_alloc(struct kmem_c if (unlikely(node != NUMA_NO_NODE && !node_isset(node, slab_nodes))) node = NUMA_NO_NODE; @@ -27,7 +26,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> goto new_slab; } redo: -@@ -2737,6 +2732,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2751,6 +2746,7 @@ static void *___slab_alloc(struct kmem_c if (!freelist) { c->page = NULL; @@ -35,7 +34,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> stat(s, DEACTIVATE_BYPASS); goto new_slab; } -@@ -2766,12 +2762,19 @@ static void *___slab_alloc(struct kmem_c +@@ -2780,12 +2776,19 @@ static void *___slab_alloc(struct kmem_c goto reread_page; } deactivate_slab(s, page, c->freelist, c); @@ -57,7 +56,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> page = c->page = slub_percpu_partial(c); slub_set_percpu_partial(c, page); local_irq_restore(flags); -@@ -2779,6 +2782,16 @@ static void *___slab_alloc(struct kmem_c +@@ -2793,6 +2796,16 @@ static void *___slab_alloc(struct kmem_c goto redo; } @@ -74,7 +73,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> freelist = get_partial(s, gfpflags, node, &page); if (freelist) { c->page = page; -@@ -2811,15 +2824,18 @@ static void *___slab_alloc(struct kmem_c +@@ -2825,15 +2838,18 @@ static void *___slab_alloc(struct kmem_c check_new_page: if (kmem_cache_debug(s)) { diff --git a/patches/0015-mm-slub-restore-irqs-around-calling-new_slab.patch b/patches/0014-mm-slub-restore-irqs-around-calling-new_slab.patch index 09feac10c056..d1801d8cd0a4 100644 --- a/patches/0015-mm-slub-restore-irqs-around-calling-new_slab.patch +++ b/patches/0014-mm-slub-restore-irqs-around-calling-new_slab.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:12 +0200 -Subject: [PATCH 15/35] mm, slub: restore irqs around calling new_slab() +Date: Mon, 10 May 2021 16:30:01 +0200 +Subject: [PATCH 14/33] mm, slub: restore irqs around calling new_slab() allocate_slab() currently re-enables irqs before calling to the page allocator. It depends on gfpflags_allow_blocking() to determine if it's safe to do so. @@ -8,14 +8,13 @@ Now we can instead simply restore irq before calling it through new_slab(). The other caller early_kmem_cache_node_alloc() is unaffected by this. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -1795,9 +1795,6 @@ static struct page *allocate_slab(struct +@@ -1809,9 +1809,6 @@ static struct page *allocate_slab(struct flags &= gfp_allowed_mask; @@ -25,7 +24,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> flags |= s->allocflags; /* -@@ -1856,8 +1853,6 @@ static struct page *allocate_slab(struct +@@ -1870,8 +1867,6 @@ static struct page *allocate_slab(struct page->frozen = 1; out: @@ -34,7 +33,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (!page) return NULL; -@@ -2798,16 +2793,17 @@ static void *___slab_alloc(struct kmem_c +@@ -2812,16 +2807,17 @@ static void *___slab_alloc(struct kmem_c goto check_new_page; } diff --git a/patches/0016-mm-slub-validate-slab-from-partial-list-or-page-allo.patch b/patches/0015-mm-slub-validate-slab-from-partial-list-or-page-allo.patch index 3926e84c00e7..4373ad8a486b 100644 --- a/patches/0016-mm-slub-validate-slab-from-partial-list-or-page-allo.patch +++ b/patches/0015-mm-slub-validate-slab-from-partial-list-or-page-allo.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:13 +0200 -Subject: [PATCH 16/35] mm, slub: validate slab from partial list or page +Date: Tue, 11 May 2021 16:37:51 +0200 +Subject: [PATCH 15/33] mm, slub: validate slab from partial list or page allocator before making it cpu slab When we obtain a new slab page from node partial list or page allocator, we @@ -12,14 +12,13 @@ so that the checks are done first, and kmem_cache_cpu.page assignment only after they pass. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 17 +++++++++-------- 1 file changed, 9 insertions(+), 8 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2788,10 +2788,8 @@ static void *___slab_alloc(struct kmem_c +@@ -2802,10 +2802,8 @@ static void *___slab_alloc(struct kmem_c lockdep_assert_irqs_disabled(); freelist = get_partial(s, gfpflags, node, &page); @@ -31,7 +30,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> local_irq_restore(flags); put_cpu_ptr(s->cpu_slab); -@@ -2804,9 +2802,6 @@ static void *___slab_alloc(struct kmem_c +@@ -2818,9 +2816,6 @@ static void *___slab_alloc(struct kmem_c } local_irq_save(flags); @@ -41,7 +40,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * No other reference to the page yet so we can * muck around with it freely without cmpxchg -@@ -2815,14 +2810,12 @@ static void *___slab_alloc(struct kmem_c +@@ -2829,14 +2824,12 @@ static void *___slab_alloc(struct kmem_c page->freelist = NULL; stat(s, ALLOC_SLAB); @@ -56,7 +55,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> local_irq_restore(flags); goto new_slab; } else { -@@ -2841,10 +2834,18 @@ static void *___slab_alloc(struct kmem_c +@@ -2855,10 +2848,18 @@ static void *___slab_alloc(struct kmem_c */ goto return_single; diff --git a/patches/0017-mm-slub-check-new-pages-with-restored-irqs.patch b/patches/0016-mm-slub-check-new-pages-with-restored-irqs.patch index 356f13571bc5..72776818ac30 100644 --- a/patches/0017-mm-slub-check-new-pages-with-restored-irqs.patch +++ b/patches/0016-mm-slub-check-new-pages-with-restored-irqs.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:14 +0200 -Subject: [PATCH 17/35] mm, slub: check new pages with restored irqs +Date: Tue, 11 May 2021 16:56:09 +0200 +Subject: [PATCH 16/33] mm, slub: check new pages with restored irqs Building on top of the previous patch, re-enable irqs before checking new pages. alloc_debug_processing() is now called with enabled irqs so we need to @@ -8,14 +8,13 @@ remove VM_BUG_ON(!irqs_disabled()); in check_slab() - there doesn't seem to be a need for it anyway. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 8 +++----- 1 file changed, 3 insertions(+), 5 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -995,8 +995,6 @@ static int check_slab(struct kmem_cache +@@ -1009,8 +1009,6 @@ static int check_slab(struct kmem_cache { int maxobj; @@ -24,7 +23,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (!PageSlab(page)) { slab_err(s, page, "Not a valid slab page"); return 0; -@@ -2788,10 +2786,10 @@ static void *___slab_alloc(struct kmem_c +@@ -2802,10 +2800,10 @@ static void *___slab_alloc(struct kmem_c lockdep_assert_irqs_disabled(); freelist = get_partial(s, gfpflags, node, &page); @@ -36,7 +35,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> put_cpu_ptr(s->cpu_slab); page = new_slab(s, gfpflags, node); c = get_cpu_ptr(s->cpu_slab); -@@ -2801,7 +2799,6 @@ static void *___slab_alloc(struct kmem_c +@@ -2815,7 +2813,6 @@ static void *___slab_alloc(struct kmem_c return NULL; } @@ -44,7 +43,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * No other reference to the page yet so we can * muck around with it freely without cmpxchg -@@ -2816,7 +2813,6 @@ static void *___slab_alloc(struct kmem_c +@@ -2830,7 +2827,6 @@ static void *___slab_alloc(struct kmem_c if (kmem_cache_debug(s)) { if (!alloc_debug_processing(s, page, freelist, addr)) { /* Slab failed checks. Next slab needed */ @@ -52,7 +51,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> goto new_slab; } else { /* -@@ -2834,6 +2830,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2848,6 +2844,7 @@ static void *___slab_alloc(struct kmem_c */ goto return_single; @@ -60,7 +59,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (unlikely(c->page)) flush_slab(s, c); c->page = page; -@@ -2842,6 +2839,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2856,6 +2853,7 @@ static void *___slab_alloc(struct kmem_c return_single: diff --git a/patches/0018-mm-slub-stop-disabling-irqs-around-get_partial.patch b/patches/0017-mm-slub-stop-disabling-irqs-around-get_partial.patch index 147016573331..8a837152186c 100644 --- a/patches/0018-mm-slub-stop-disabling-irqs-around-get_partial.patch +++ b/patches/0017-mm-slub-stop-disabling-irqs-around-get_partial.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:15 +0200 -Subject: [PATCH 18/35] mm, slub: stop disabling irqs around get_partial() +Date: Tue, 11 May 2021 17:45:26 +0200 +Subject: [PATCH 17/33] mm, slub: stop disabling irqs around get_partial() The function get_partial() does not need to have irqs disabled as a whole. It's sufficient to convert spin_lock operations to their irq saving/restoring @@ -10,14 +10,13 @@ As a result, it's now possible to reach the page allocator from the slab allocator without disabling and re-enabling interrupts on the way. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 22 ++++++++-------------- 1 file changed, 8 insertions(+), 14 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -1996,11 +1996,12 @@ static inline bool pfmemalloc_match(stru +@@ -2010,11 +2010,12 @@ static inline bool pfmemalloc_match(stru * Try to allocate a partial slab from a specific node. */ static void *get_partial_node(struct kmem_cache *s, struct kmem_cache_node *n, @@ -31,7 +30,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> int objects; /* -@@ -2012,11 +2013,11 @@ static void *get_partial_node(struct kme +@@ -2026,11 +2027,11 @@ static void *get_partial_node(struct kme if (!n || !n->nr_partial) return NULL; @@ -45,7 +44,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> continue; t = acquire_slab(s, n, page, object == NULL, &objects); -@@ -2037,7 +2038,7 @@ static void *get_partial_node(struct kme +@@ -2051,7 +2052,7 @@ static void *get_partial_node(struct kme break; } @@ -54,7 +53,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> return object; } -@@ -2765,8 +2766,10 @@ static void *___slab_alloc(struct kmem_c +@@ -2779,8 +2780,10 @@ static void *___slab_alloc(struct kmem_c local_irq_restore(flags); goto reread_page; } @@ -66,7 +65,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> page = c->page = slub_percpu_partial(c); slub_set_percpu_partial(c, page); -@@ -2775,18 +2778,9 @@ static void *___slab_alloc(struct kmem_c +@@ -2789,18 +2792,9 @@ static void *___slab_alloc(struct kmem_c goto redo; } diff --git a/patches/0019-mm-slub-move-reset-of-c-page-and-freelist-out-of-dea.patch b/patches/0018-mm-slub-move-reset-of-c-page-and-freelist-out-of-dea.patch index a7beb833a7c1..71dc9fbb2803 100644 --- a/patches/0019-mm-slub-move-reset-of-c-page-and-freelist-out-of-dea.patch +++ b/patches/0018-mm-slub-move-reset-of-c-page-and-freelist-out-of-dea.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:16 +0200 -Subject: [PATCH 19/35] mm, slub: move reset of c->page and freelist out of +Date: Wed, 12 May 2021 13:53:34 +0200 +Subject: [PATCH 18/33] mm, slub: move reset of c->page and freelist out of deactivate_slab() deactivate_slab() removes the cpu slab by merging the cpu freelist with slab's @@ -15,14 +15,13 @@ to assign kmem_cache_cpu.page before deactivation or care if somebody preempted us and assigned a different page to our kmem_cache_cpu in the process. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 31 ++++++++++++++++++------------- 1 file changed, 18 insertions(+), 13 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2195,10 +2195,13 @@ static void init_kmem_cache_cpus(struct +@@ -2209,10 +2209,13 @@ static void init_kmem_cache_cpus(struct } /* @@ -38,7 +37,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> { enum slab_modes { M_NONE, M_PARTIAL, M_FULL, M_FREE }; struct kmem_cache_node *n = get_node(s, page_to_nid(page)); -@@ -2327,9 +2330,6 @@ static void deactivate_slab(struct kmem_ +@@ -2341,9 +2344,6 @@ static void deactivate_slab(struct kmem_ discard_slab(s, page); stat(s, FREE_SLAB); } @@ -48,7 +47,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } /* -@@ -2454,10 +2454,16 @@ static void put_cpu_partial(struct kmem_ +@@ -2468,10 +2468,16 @@ static void put_cpu_partial(struct kmem_ static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c) { @@ -67,7 +66,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } /* -@@ -2755,7 +2761,10 @@ static void *___slab_alloc(struct kmem_c +@@ -2769,7 +2775,10 @@ static void *___slab_alloc(struct kmem_c local_irq_restore(flags); goto reread_page; } @@ -79,7 +78,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> local_irq_restore(flags); new_slab: -@@ -2834,11 +2843,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2848,11 +2857,7 @@ static void *___slab_alloc(struct kmem_c return_single: local_irq_save(flags); diff --git a/patches/0020-mm-slub-make-locking-in-deactivate_slab-irq-safe.patch b/patches/0019-mm-slub-make-locking-in-deactivate_slab-irq-safe.patch index 678355c7d274..3373cebc4aa4 100644 --- a/patches/0020-mm-slub-make-locking-in-deactivate_slab-irq-safe.patch +++ b/patches/0019-mm-slub-make-locking-in-deactivate_slab-irq-safe.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:17 +0200 -Subject: [PATCH 20/35] mm, slub: make locking in deactivate_slab() irq-safe +Date: Wed, 12 May 2021 13:59:58 +0200 +Subject: [PATCH 19/33] mm, slub: make locking in deactivate_slab() irq-safe dectivate_slab() now no longer touches the kmem_cache_cpu structure, so it will be possible to call it with irqs enabled. Just convert the spin_lock calls to @@ -10,14 +10,13 @@ Note we now have to use cmpxchg_double_slab() for irq-safe slab_lock(), because in some situations we don't take the list_lock, which would disable irqs. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2209,6 +2209,7 @@ static void deactivate_slab(struct kmem_ +@@ -2223,6 +2223,7 @@ static void deactivate_slab(struct kmem_ enum slab_modes l = M_NONE, m = M_NONE; void *nextfree, *freelist_iter, *freelist_tail; int tail = DEACTIVATE_TO_HEAD; @@ -25,7 +24,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> struct page new; struct page old; -@@ -2284,7 +2285,7 @@ static void deactivate_slab(struct kmem_ +@@ -2298,7 +2299,7 @@ static void deactivate_slab(struct kmem_ * that acquire_slab() will see a slab page that * is frozen */ @@ -34,7 +33,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } } else { m = M_FULL; -@@ -2295,7 +2296,7 @@ static void deactivate_slab(struct kmem_ +@@ -2309,7 +2310,7 @@ static void deactivate_slab(struct kmem_ * slabs from diagnostic functions will not see * any frozen slabs. */ @@ -43,7 +42,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } } -@@ -2312,14 +2313,14 @@ static void deactivate_slab(struct kmem_ +@@ -2326,14 +2327,14 @@ static void deactivate_slab(struct kmem_ } l = m; diff --git a/patches/0021-mm-slub-call-deactivate_slab-without-disabling-irqs.patch b/patches/0020-mm-slub-call-deactivate_slab-without-disabling-irqs.patch index 576b90ba7e59..90b6157b1bde 100644 --- a/patches/0021-mm-slub-call-deactivate_slab-without-disabling-irqs.patch +++ b/patches/0020-mm-slub-call-deactivate_slab-without-disabling-irqs.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:18 +0200 -Subject: [PATCH 21/35] mm, slub: call deactivate_slab() without disabling irqs +Date: Wed, 12 May 2021 14:04:43 +0200 +Subject: [PATCH 20/33] mm, slub: call deactivate_slab() without disabling irqs The function is now safe to be called with irqs enabled, so move the calls outside of irq disabled sections. @@ -16,14 +16,13 @@ irqs anyway, and slub_cpu_dead() which will be dealt with in the following patch. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 24 +++++++++++++++++++----- 1 file changed, 19 insertions(+), 5 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2765,8 +2765,8 @@ static void *___slab_alloc(struct kmem_c +@@ -2779,8 +2779,8 @@ static void *___slab_alloc(struct kmem_c freelist = c->freelist; c->page = NULL; c->freelist = NULL; @@ -33,7 +32,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> new_slab: -@@ -2834,18 +2834,32 @@ static void *___slab_alloc(struct kmem_c +@@ -2848,18 +2848,32 @@ static void *___slab_alloc(struct kmem_c */ goto return_single; diff --git a/patches/0022-mm-slub-move-irq-control-into-unfreeze_partials.patch b/patches/0021-mm-slub-move-irq-control-into-unfreeze_partials.patch index 6a79f28bb712..55d97a8a4893 100644 --- a/patches/0022-mm-slub-move-irq-control-into-unfreeze_partials.patch +++ b/patches/0021-mm-slub-move-irq-control-into-unfreeze_partials.patch @@ -1,20 +1,19 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:19 +0200 -Subject: [PATCH 22/35] mm, slub: move irq control into unfreeze_partials() +Date: Thu, 20 May 2021 14:00:03 +0200 +Subject: [PATCH 21/33] mm, slub: move irq control into unfreeze_partials() unfreeze_partials() can be optimized so that it doesn't need irqs disabled for the whole time. As the first step, move irq control into the function and remove it from the put_cpu_partial() caller. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2336,9 +2336,8 @@ static void deactivate_slab(struct kmem_ +@@ -2350,9 +2350,8 @@ static void deactivate_slab(struct kmem_ /* * Unfreeze all the cpu partial slabs. * @@ -26,7 +25,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> */ static void unfreeze_partials(struct kmem_cache *s, struct kmem_cache_cpu *c) -@@ -2346,6 +2345,9 @@ static void unfreeze_partials(struct kme +@@ -2360,6 +2359,9 @@ static void unfreeze_partials(struct kme #ifdef CONFIG_SLUB_CPU_PARTIAL struct kmem_cache_node *n = NULL, *n2 = NULL; struct page *page, *discard_page = NULL; @@ -36,7 +35,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> while ((page = slub_percpu_partial(c))) { struct page new; -@@ -2398,6 +2400,8 @@ static void unfreeze_partials(struct kme +@@ -2412,6 +2414,8 @@ static void unfreeze_partials(struct kme discard_slab(s, page); stat(s, FREE_SLAB); } @@ -45,7 +44,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> #endif /* CONFIG_SLUB_CPU_PARTIAL */ } -@@ -2425,14 +2429,11 @@ static void put_cpu_partial(struct kmem_ +@@ -2439,14 +2443,11 @@ static void put_cpu_partial(struct kmem_ pobjects = oldpage->pobjects; pages = oldpage->pages; if (drain && pobjects > slub_cpu_partial(s)) { diff --git a/patches/0023-mm-slub-discard-slabs-in-unfreeze_partials-without-i.patch b/patches/0022-mm-slub-discard-slabs-in-unfreeze_partials-without-i.patch index 58d0c6a18180..5a81f66bd24a 100644 --- a/patches/0023-mm-slub-discard-slabs-in-unfreeze_partials-without-i.patch +++ b/patches/0022-mm-slub-discard-slabs-in-unfreeze_partials-without-i.patch @@ -1,20 +1,19 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:20 +0200 -Subject: [PATCH 23/35] mm, slub: discard slabs in unfreeze_partials() without +Date: Thu, 20 May 2021 14:01:57 +0200 +Subject: [PATCH 22/33] mm, slub: discard slabs in unfreeze_partials() without irqs disabled No need for disabled irqs when discarding slabs, so restore them before discarding. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2392,6 +2392,8 @@ static void unfreeze_partials(struct kme +@@ -2406,6 +2406,8 @@ static void unfreeze_partials(struct kme if (n) spin_unlock(&n->list_lock); @@ -23,7 +22,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> while (discard_page) { page = discard_page; discard_page = discard_page->next; -@@ -2401,7 +2403,6 @@ static void unfreeze_partials(struct kme +@@ -2415,7 +2417,6 @@ static void unfreeze_partials(struct kme stat(s, FREE_SLAB); } diff --git a/patches/0024-mm-slub-detach-whole-partial-list-at-once-in-unfreez.patch b/patches/0023-mm-slub-detach-whole-partial-list-at-once-in-unfreez.patch index 4b36f4376cb4..627a8c94488e 100644 --- a/patches/0024-mm-slub-detach-whole-partial-list-at-once-in-unfreez.patch +++ b/patches/0023-mm-slub-detach-whole-partial-list-at-once-in-unfreez.patch @@ -1,20 +1,19 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:21 +0200 -Subject: [PATCH 24/35] mm, slub: detach whole partial list at once in +Date: Thu, 20 May 2021 14:18:12 +0200 +Subject: [PATCH 23/33] mm, slub: detach whole partial list at once in unfreeze_partials() Instead of iterating through the live percpu partial list, detach it from the kmem_cache_cpu at once. This is simpler and will allow further optimization. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2344,16 +2344,20 @@ static void unfreeze_partials(struct kme +@@ -2358,16 +2358,20 @@ static void unfreeze_partials(struct kme { #ifdef CONFIG_SLUB_CPU_PARTIAL struct kmem_cache_node *n = NULL, *n2 = NULL; diff --git a/patches/0025-mm-slub-separate-detaching-of-partial-list-in-unfree.patch b/patches/0024-mm-slub-separate-detaching-of-partial-list-in-unfree.patch index 363cfdd1c429..7a01091b8456 100644 --- a/patches/0025-mm-slub-separate-detaching-of-partial-list-in-unfree.patch +++ b/patches/0024-mm-slub-separate-detaching-of-partial-list-in-unfree.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:22 +0200 -Subject: [PATCH 25/35] mm, slub: separate detaching of partial list in +Date: Thu, 20 May 2021 16:39:51 +0200 +Subject: [PATCH 24/33] mm, slub: separate detaching of partial list in unfreeze_partials() from unfreezing Unfreezing partial list can be split to two phases - detaching the list from @@ -25,14 +25,13 @@ restructure the code as follows: proper calls. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 73 +++++++++++++++++++++++++++++++++++++++++++------------------- 1 file changed, 51 insertions(+), 22 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2333,25 +2333,15 @@ static void deactivate_slab(struct kmem_ +@@ -2347,25 +2347,15 @@ static void deactivate_slab(struct kmem_ } } @@ -61,7 +60,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> while (partial_page) { struct page new; struct page old; -@@ -2406,10 +2396,45 @@ static void unfreeze_partials(struct kme +@@ -2420,10 +2410,45 @@ static void unfreeze_partials(struct kme discard_slab(s, page); stat(s, FREE_SLAB); } @@ -108,7 +107,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * Put a page that was just frozen (in __slab_free|get_partial_node) into a * partial page slot if available. -@@ -2438,7 +2463,7 @@ static void put_cpu_partial(struct kmem_ +@@ -2452,7 +2477,7 @@ static void put_cpu_partial(struct kmem_ * partial array is full. Move the existing * set to the per node partial list. */ @@ -117,7 +116,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> oldpage = NULL; pobjects = 0; pages = 0; -@@ -2473,11 +2498,6 @@ static inline void flush_slab(struct kme +@@ -2487,11 +2512,6 @@ static inline void flush_slab(struct kme stat(s, CPUSLAB_FLUSH); } @@ -129,7 +128,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> static inline void __flush_cpu_slab(struct kmem_cache *s, int cpu) { struct kmem_cache_cpu *c = per_cpu_ptr(s->cpu_slab, cpu); -@@ -2485,14 +2505,23 @@ static inline void __flush_cpu_slab(stru +@@ -2499,14 +2519,23 @@ static inline void __flush_cpu_slab(stru if (c->page) flush_slab(s, c); diff --git a/patches/0026-mm-slub-only-disable-irq-with-spin_lock-in-__unfreez.patch b/patches/0025-mm-slub-only-disable-irq-with-spin_lock-in-__unfreez.patch index 6f52a063a54b..8d8f92eb7a96 100644 --- a/patches/0026-mm-slub-only-disable-irq-with-spin_lock-in-__unfreez.patch +++ b/patches/0025-mm-slub-only-disable-irq-with-spin_lock-in-__unfreez.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:23 +0200 -Subject: [PATCH 26/35] mm, slub: only disable irq with spin_lock in +Date: Fri, 21 May 2021 01:16:54 +0200 +Subject: [PATCH 25/33] mm, slub: only disable irq with spin_lock in __unfreeze_partials() __unfreeze_partials() no longer needs to have irqs disabled, except for making @@ -8,14 +8,13 @@ the spin_lock operations irq-safe, so convert the spin_locks operations and remove the separate irq handling. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 12 ++++-------- 1 file changed, 4 insertions(+), 8 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2338,9 +2338,7 @@ static void __unfreeze_partials(struct k +@@ -2352,9 +2352,7 @@ static void __unfreeze_partials(struct k { struct kmem_cache_node *n = NULL, *n2 = NULL; struct page *page, *discard_page = NULL; @@ -26,7 +25,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> while (partial_page) { struct page new; -@@ -2352,10 +2350,10 @@ static void __unfreeze_partials(struct k +@@ -2366,10 +2364,10 @@ static void __unfreeze_partials(struct k n2 = get_node(s, page_to_nid(page)); if (n != n2) { if (n) @@ -39,7 +38,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } do { -@@ -2384,9 +2382,7 @@ static void __unfreeze_partials(struct k +@@ -2398,9 +2396,7 @@ static void __unfreeze_partials(struct k } if (n) diff --git a/patches/0027-mm-slub-don-t-disable-irqs-in-slub_cpu_dead.patch b/patches/0026-mm-slub-don-t-disable-irqs-in-slub_cpu_dead.patch index cf715091134f..ecc53e05a8e6 100644 --- a/patches/0027-mm-slub-don-t-disable-irqs-in-slub_cpu_dead.patch +++ b/patches/0026-mm-slub-don-t-disable-irqs-in-slub_cpu_dead.patch @@ -1,19 +1,18 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:24 +0200 -Subject: [PATCH 27/35] mm, slub: don't disable irqs in slub_cpu_dead() +Date: Fri, 21 May 2021 01:48:56 +0200 +Subject: [PATCH 26/33] mm, slub: don't disable irqs in slub_cpu_dead() slub_cpu_dead() cleans up for an offlined cpu from another cpu and calls only functions that are now irq safe, so we don't need to disable irqs anymore. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2540,14 +2540,10 @@ static void flush_all(struct kmem_cache +@@ -2554,14 +2554,10 @@ static void flush_all(struct kmem_cache static int slub_cpu_dead(unsigned int cpu) { struct kmem_cache *s; diff --git a/patches/0027-mm-slab-split-out-the-cpu-offline-variant-of-flush_s.patch b/patches/0027-mm-slab-split-out-the-cpu-offline-variant-of-flush_s.patch new file mode 100644 index 000000000000..d4a58b39704b --- /dev/null +++ b/patches/0027-mm-slab-split-out-the-cpu-offline-variant-of-flush_s.patch @@ -0,0 +1,44 @@ +From: Vlastimil Babka <vbabka@suse.cz> +Date: Thu, 3 Jun 2021 19:17:42 +0200 +Subject: [PATCH 27/33] mm, slab: split out the cpu offline variant of + flush_slab() + +flush_slab() is called either as part IPI handler on given live cpu, or as a +cleanup on behalf of another cpu that went offline. The first case needs to +protect updating the kmem_cache_cpu fields with disabled irqs. Currently the +whole call happens with irqs disabled by the IPI handler, but the following +patch will change from IPI to workqueue, and flush_slab() will have to disable +irqs (to be replaced with a local lock later) in the critical part. + +To prepare for this change, replace the call to flush_slab() for the dead cpu +handling with an opencoded variant that will not disable irqs nor take a local +lock. + +Suggested-by: Mike Galbraith <efault@gmx.de> +Signed-off-by: Vlastimil Babka <vbabka@suse.cz> +--- + mm/slub.c | 12 ++++++++++-- + 1 file changed, 10 insertions(+), 2 deletions(-) + +--- a/mm/slub.c ++++ b/mm/slub.c +@@ -2511,9 +2511,17 @@ static inline void flush_slab(struct kme + static inline void __flush_cpu_slab(struct kmem_cache *s, int cpu) + { + struct kmem_cache_cpu *c = per_cpu_ptr(s->cpu_slab, cpu); ++ void *freelist = c->freelist; ++ struct page *page = c->page; + +- if (c->page) +- flush_slab(s, c); ++ c->page = NULL; ++ c->freelist = NULL; ++ c->tid = next_tid(c->tid); ++ ++ if (page) { ++ deactivate_slab(s, page, freelist); ++ stat(s, CPUSLAB_FLUSH); ++ } + + unfreeze_partials_cpu(s, c); + } diff --git a/patches/0028-mm-slab-make-flush_slab-possible-to-call-with-irqs-e.patch b/patches/0028-mm-slab-make-flush_slab-possible-to-call-with-irqs-e.patch deleted file mode 100644 index dd475caaf163..000000000000 --- a/patches/0028-mm-slab-make-flush_slab-possible-to-call-with-irqs-e.patch +++ /dev/null @@ -1,69 +0,0 @@ -From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:25 +0200 -Subject: [PATCH 28/35] mm, slab: make flush_slab() possible to call with irqs - enabled - -Currently flush_slab() is always called with disabled IRQs if it's needed, but -the following patches will change that, so add a parameter to control IRQ -disabling within the function, which only protects the kmem_cache_cpu -manipulation and not the call to deactivate_slab() which doesn't need it. - -Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> ---- - mm/slub.c | 24 ++++++++++++++++++------ - 1 file changed, 18 insertions(+), 6 deletions(-) - ---- a/mm/slub.c -+++ b/mm/slub.c -@@ -2480,16 +2480,28 @@ static void put_cpu_partial(struct kmem_ - #endif /* CONFIG_SLUB_CPU_PARTIAL */ - } - --static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c) -+static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c, -+ bool lock) - { -- void *freelist = c->freelist; -- struct page *page = c->page; -+ unsigned long flags; -+ void *freelist; -+ struct page *page; -+ -+ if (lock) -+ local_irq_save(flags); -+ -+ freelist = c->freelist; -+ page = c->page; - - c->page = NULL; - c->freelist = NULL; - c->tid = next_tid(c->tid); - -- deactivate_slab(s, page, freelist); -+ if (lock) -+ local_irq_restore(flags); -+ -+ if (page) -+ deactivate_slab(s, page, freelist); - - stat(s, CPUSLAB_FLUSH); - } -@@ -2499,7 +2511,7 @@ static inline void __flush_cpu_slab(stru - struct kmem_cache_cpu *c = per_cpu_ptr(s->cpu_slab, cpu); - - if (c->page) -- flush_slab(s, c); -+ flush_slab(s, c, false); - - unfreeze_partials_cpu(s, c); - } -@@ -2515,7 +2527,7 @@ static void flush_cpu_slab(void *d) - struct kmem_cache_cpu *c = this_cpu_ptr(s->cpu_slab); - - if (c->page) -- flush_slab(s, c); -+ flush_slab(s, c, false); - - unfreeze_partials(s); - } diff --git a/patches/0029-mm-slub-Move-flush_cpu_slab-invocations-__free_slab-.patch b/patches/0028-mm-slub-move-flush_cpu_slab-invocations-__free_slab-.patch index 641577dcc7e7..78b29d9aeabd 100644 --- a/patches/0029-mm-slub-Move-flush_cpu_slab-invocations-__free_slab-.patch +++ b/patches/0028-mm-slub-move-flush_cpu_slab-invocations-__free_slab-.patch @@ -1,6 +1,6 @@ From: Sebastian Andrzej Siewior <bigeasy@linutronix.de> -Date: Thu, 29 Jul 2021 15:21:26 +0200 -Subject: [PATCH 29/35] mm: slub: Move flush_cpu_slab() invocations +Date: Fri, 26 Feb 2021 17:11:55 +0100 +Subject: [PATCH 28/33] mm: slub: move flush_cpu_slab() invocations __free_slab() invocations out of IRQ context flush_all() flushes a specific SLAB cache on each CPU (where the cache @@ -10,14 +10,17 @@ within IPI handler and is problematic for PREEMPT_RT. The flush operation is not a frequent operation or a hot path. The per-CPU flush operation can be moved to within a workqueue. +Because a workqueue handler, unlike IPI handler, does not disable irqs, +flush_slab() now has to disable them for working with the kmem_cache_cpu +fields. deactivate_slab() is safe to call with irqs enabled. + [vbabka@suse.cz: adapt to new SLUB changes] Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slab_common.c | 2 + - mm/slub.c | 79 +++++++++++++++++++++++++++++++++++++++++++++---------- - 2 files changed, 68 insertions(+), 13 deletions(-) + mm/slub.c | 94 +++++++++++++++++++++++++++++++++++++++++++++---------- + 2 files changed, 80 insertions(+), 16 deletions(-) --- a/mm/slab_common.c +++ b/mm/slab_common.c @@ -39,7 +42,37 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- a/mm/slub.c +++ b/mm/slub.c -@@ -2516,33 +2516,79 @@ static inline void __flush_cpu_slab(stru +@@ -2496,16 +2496,25 @@ static void put_cpu_partial(struct kmem_ + + static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c) + { +- void *freelist = c->freelist; +- struct page *page = c->page; ++ unsigned long flags; ++ struct page *page; ++ void *freelist; ++ ++ local_irq_save(flags); ++ ++ page = c->page; ++ freelist = c->freelist; + + c->page = NULL; + c->freelist = NULL; + c->tid = next_tid(c->tid); + +- deactivate_slab(s, page, freelist); ++ local_irq_restore(flags); + +- stat(s, CPUSLAB_FLUSH); ++ if (page) { ++ deactivate_slab(s, page, freelist); ++ stat(s, CPUSLAB_FLUSH); ++ } + } + + static inline void __flush_cpu_slab(struct kmem_cache *s, int cpu) +@@ -2526,15 +2535,27 @@ static inline void __flush_cpu_slab(stru unfreeze_partials_cpu(s, c); } @@ -70,9 +103,8 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> + c = this_cpu_ptr(s->cpu_slab); if (c->page) -- flush_slab(s, c, false); -+ flush_slab(s, c, true); - + flush_slab(s, c); +@@ -2542,17 +2563,51 @@ static void flush_cpu_slab(void *d) unfreeze_partials(s); } @@ -127,7 +159,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } /* -@@ -4087,7 +4133,7 @@ int __kmem_cache_shutdown(struct kmem_ca +@@ -4097,7 +4152,7 @@ int __kmem_cache_shutdown(struct kmem_ca int node; struct kmem_cache_node *n; @@ -136,7 +168,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* Attempt to free all objects */ for_each_kmem_cache_node(s, node, n) { free_partial(s, n); -@@ -4363,7 +4409,7 @@ EXPORT_SYMBOL(kfree); +@@ -4373,7 +4428,7 @@ EXPORT_SYMBOL(kfree); * being allocated from last increasing the chance that the last objects * are freed in them. */ @@ -145,7 +177,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> { int node; int i; -@@ -4375,7 +4421,6 @@ int __kmem_cache_shrink(struct kmem_cach +@@ -4385,7 +4440,6 @@ int __kmem_cache_shrink(struct kmem_cach unsigned long flags; int ret = 0; @@ -153,7 +185,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> for_each_kmem_cache_node(s, node, n) { INIT_LIST_HEAD(&discard); for (i = 0; i < SHRINK_PROMOTE_MAX; i++) -@@ -4425,13 +4470,21 @@ int __kmem_cache_shrink(struct kmem_cach +@@ -4435,13 +4489,21 @@ int __kmem_cache_shrink(struct kmem_cach return ret; } diff --git a/patches/0030-mm-slub-Make-object_map_lock-a-raw_spinlock_t.patch b/patches/0029-mm-slub-make-object_map_lock-a-raw_spinlock_t.patch index 0cdb2c71066e..a058bfd30507 100644 --- a/patches/0030-mm-slub-Make-object_map_lock-a-raw_spinlock_t.patch +++ b/patches/0029-mm-slub-make-object_map_lock-a-raw_spinlock_t.patch @@ -1,6 +1,6 @@ From: Sebastian Andrzej Siewior <bigeasy@linutronix.de> -Date: Thu, 29 Jul 2021 15:21:27 +0200 -Subject: [PATCH 30/35] mm: slub: Make object_map_lock a raw_spinlock_t +Date: Thu, 16 Jul 2020 18:47:50 +0200 +Subject: [PATCH 29/33] mm: slub: make object_map_lock a raw_spinlock_t The variable object_map is protected by object_map_lock. The lock is always acquired in debug code and within already atomic context @@ -9,14 +9,13 @@ Make object_map_lock a raw_spinlock_t. Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -438,7 +438,7 @@ static inline bool cmpxchg_double_slab(s +@@ -452,7 +452,7 @@ static inline bool cmpxchg_double_slab(s #ifdef CONFIG_SLUB_DEBUG static unsigned long object_map[BITS_TO_LONGS(MAX_OBJS_PER_PAGE)]; @@ -25,7 +24,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> static void __fill_map(unsigned long *obj_map, struct kmem_cache *s, struct page *page) -@@ -483,7 +483,7 @@ static unsigned long *get_map(struct kme +@@ -497,7 +497,7 @@ static unsigned long *get_map(struct kme { VM_BUG_ON(!irqs_disabled()); @@ -34,7 +33,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> __fill_map(object_map, s, page); -@@ -493,7 +493,7 @@ static unsigned long *get_map(struct kme +@@ -507,7 +507,7 @@ static unsigned long *get_map(struct kme static void put_map(unsigned long *map) __releases(&object_map_lock) { VM_BUG_ON(map != object_map); diff --git a/patches/0030-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch b/patches/0030-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch new file mode 100644 index 000000000000..9453152ed8f2 --- /dev/null +++ b/patches/0030-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch @@ -0,0 +1,186 @@ +From: Vlastimil Babka <vbabka@suse.cz> +Date: Fri, 4 Jun 2021 12:55:55 +0200 +Subject: [PATCH 30/33] mm, slub: make slab_lock() disable irqs with PREEMPT_RT + +We need to disable irqs around slab_lock() (a bit spinlock) to make it +irq-safe. Most calls to slab_lock() are nested under spin_lock_irqsave() which +doesn't disable irqs on PREEMPT_RT, so add explicit disabling with PREEMPT_RT. +The exception is cmpxchg_double_slab() which already disables irqs, so use a +__slab_[un]lock() variant without irq disable there. + +slab_[un]lock() thus needs a flags pointer parameter, which is unused on !RT. +free_debug_processing() now has two flags variables, which looks odd, but only +one is actually used - the one used in spin_lock_irqsave() on !RT and the one +used in slab_lock() on RT. + +As a result, __cmpxchg_double_slab() and cmpxchg_double_slab() become +effectively identical on RT, as both will disable irqs, which is necessary on +RT as most callers of this function also rely on irqsaving lock operations. +Thus, assert that irqs are already disabled in __cmpxchg_double_slab() only on +!RT and also change the VM_BUG_ON assertion to the more standard lockdep_assert +one. + +Signed-off-by: Vlastimil Babka <vbabka@suse.cz> +--- + mm/slub.c | 58 +++++++++++++++++++++++++++++++++++++++++----------------- + 1 file changed, 41 insertions(+), 17 deletions(-) + +--- a/mm/slub.c ++++ b/mm/slub.c +@@ -359,25 +359,44 @@ static inline unsigned int oo_objects(st + /* + * Per slab locking using the pagelock + */ +-static __always_inline void slab_lock(struct page *page) ++static __always_inline void __slab_lock(struct page *page) + { + VM_BUG_ON_PAGE(PageTail(page), page); + bit_spin_lock(PG_locked, &page->flags); + } + +-static __always_inline void slab_unlock(struct page *page) ++static __always_inline void __slab_unlock(struct page *page) + { + VM_BUG_ON_PAGE(PageTail(page), page); + __bit_spin_unlock(PG_locked, &page->flags); + } + +-/* Interrupts must be disabled (for the fallback code to work right) */ ++static __always_inline void slab_lock(struct page *page, unsigned long *flags) ++{ ++ if (IS_ENABLED(CONFIG_PREEMPT_RT)) ++ local_irq_save(*flags); ++ __slab_lock(page); ++} ++ ++static __always_inline void slab_unlock(struct page *page, unsigned long *flags) ++{ ++ __slab_unlock(page); ++ if (IS_ENABLED(CONFIG_PREEMPT_RT)) ++ local_irq_restore(*flags); ++} ++ ++/* ++ * Interrupts must be disabled (for the fallback code to work right), typically ++ * by an _irqsave() lock variant. Except on PREEMPT_RT where locks are different ++ * so we disable interrupts as part of slab_[un]lock(). ++ */ + static inline bool __cmpxchg_double_slab(struct kmem_cache *s, struct page *page, + void *freelist_old, unsigned long counters_old, + void *freelist_new, unsigned long counters_new, + const char *n) + { +- VM_BUG_ON(!irqs_disabled()); ++ if (!IS_ENABLED(CONFIG_PREEMPT_RT)) ++ lockdep_assert_irqs_disabled(); + #if defined(CONFIG_HAVE_CMPXCHG_DOUBLE) && \ + defined(CONFIG_HAVE_ALIGNED_STRUCT_PAGE) + if (s->flags & __CMPXCHG_DOUBLE) { +@@ -388,15 +407,18 @@ static inline bool __cmpxchg_double_slab + } else + #endif + { +- slab_lock(page); ++ /* init to 0 to prevent spurious warnings */ ++ unsigned long flags = 0; ++ ++ slab_lock(page, &flags); + if (page->freelist == freelist_old && + page->counters == counters_old) { + page->freelist = freelist_new; + page->counters = counters_new; +- slab_unlock(page); ++ slab_unlock(page, &flags); + return true; + } +- slab_unlock(page); ++ slab_unlock(page, &flags); + } + + cpu_relax(); +@@ -427,16 +449,16 @@ static inline bool cmpxchg_double_slab(s + unsigned long flags; + + local_irq_save(flags); +- slab_lock(page); ++ __slab_lock(page); + if (page->freelist == freelist_old && + page->counters == counters_old) { + page->freelist = freelist_new; + page->counters = counters_new; +- slab_unlock(page); ++ __slab_unlock(page); + local_irq_restore(flags); + return true; + } +- slab_unlock(page); ++ __slab_unlock(page); + local_irq_restore(flags); + } + +@@ -1269,11 +1291,11 @@ static noinline int free_debug_processin + struct kmem_cache_node *n = get_node(s, page_to_nid(page)); + void *object = head; + int cnt = 0; +- unsigned long flags; ++ unsigned long flags, flags2; + int ret = 0; + + spin_lock_irqsave(&n->list_lock, flags); +- slab_lock(page); ++ slab_lock(page, &flags2); + + if (s->flags & SLAB_CONSISTENCY_CHECKS) { + if (!check_slab(s, page)) +@@ -1306,7 +1328,7 @@ static noinline int free_debug_processin + slab_err(s, page, "Bulk freelist count(%d) invalid(%d)\n", + bulk_cnt, cnt); + +- slab_unlock(page); ++ slab_unlock(page, &flags2); + spin_unlock_irqrestore(&n->list_lock, flags); + if (!ret) + slab_fix(s, "Object at 0x%p not freed", object); +@@ -4087,11 +4109,12 @@ static void list_slab_objects(struct kme + { + #ifdef CONFIG_SLUB_DEBUG + void *addr = page_address(page); ++ unsigned long flags; + unsigned long *map; + void *p; + + slab_err(s, page, text, s->name); +- slab_lock(page); ++ slab_lock(page, &flags); + + map = get_map(s, page); + for_each_object(p, s, addr, page->objects) { +@@ -4102,7 +4125,7 @@ static void list_slab_objects(struct kme + } + } + put_map(map); +- slab_unlock(page); ++ slab_unlock(page, &flags); + #endif + } + +@@ -4834,8 +4857,9 @@ static void validate_slab(struct kmem_ca + { + void *p; + void *addr = page_address(page); ++ unsigned long flags; + +- slab_lock(page); ++ slab_lock(page, &flags); + + if (!check_slab(s, page) || !on_freelist(s, page, NULL)) + goto unlock; +@@ -4850,7 +4874,7 @@ static void validate_slab(struct kmem_ca + break; + } + unlock: +- slab_unlock(page); ++ slab_unlock(page, &flags); + } + + static int validate_slab_node(struct kmem_cache *s, diff --git a/patches/0031-mm-slub-optionally-save-restore-irqs-in-slab_-un-loc.patch b/patches/0031-mm-slub-optionally-save-restore-irqs-in-slab_-un-loc.patch deleted file mode 100644 index e80e03924e66..000000000000 --- a/patches/0031-mm-slub-optionally-save-restore-irqs-in-slab_-un-loc.patch +++ /dev/null @@ -1,150 +0,0 @@ -From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:28 +0200 -Subject: [PATCH 31/35] mm, slub: optionally save/restore irqs in - slab_[un]lock()/ - -For PREEMPT_RT we will need to disable irqs for this bit spinlock. As a -preparation, add a flags parameter, and an internal version that takes -additional bool parameter to control irq saving/restoring (the flags -parameter is compile-time unused if the bool is a constant false). - -Convert ___cmpxchg_double_slab(), which also comes with the same bool -parameter, to use the internal version. - -Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> ---- - mm/slub.c | 52 +++++++++++++++++++++++++++++++++------------------- - 1 file changed, 33 insertions(+), 19 deletions(-) - ---- a/mm/slub.c -+++ b/mm/slub.c -@@ -359,16 +359,33 @@ static inline unsigned int oo_objects(st - /* - * Per slab locking using the pagelock - */ --static __always_inline void slab_lock(struct page *page) -+static __always_inline void -+__slab_lock(struct page *page, unsigned long *flags, bool disable_irqs) - { - VM_BUG_ON_PAGE(PageTail(page), page); -+ if (disable_irqs) -+ local_irq_save(*flags); - bit_spin_lock(PG_locked, &page->flags); - } - --static __always_inline void slab_unlock(struct page *page) -+static __always_inline void -+__slab_unlock(struct page *page, unsigned long *flags, bool disable_irqs) - { - VM_BUG_ON_PAGE(PageTail(page), page); - __bit_spin_unlock(PG_locked, &page->flags); -+ if (disable_irqs) -+ local_irq_restore(*flags); -+} -+ -+static __always_inline void -+slab_lock(struct page *page, unsigned long *flags) -+{ -+ __slab_lock(page, flags, false); -+} -+ -+static __always_inline void slab_unlock(struct page *page, unsigned long *flags) -+{ -+ __slab_unlock(page, flags, false); - } - - static inline bool ___cmpxchg_double_slab(struct kmem_cache *s, struct page *page, -@@ -388,23 +405,18 @@ static inline bool ___cmpxchg_double_sla - } else - #endif - { -- unsigned long flags; -+ /* init to 0 to prevent spurious warnings */ -+ unsigned long flags = 0; - -- if (disable_irqs) -- local_irq_save(flags); -- slab_lock(page); -+ __slab_lock(page, &flags, disable_irqs); - if (page->freelist == freelist_old && - page->counters == counters_old) { - page->freelist = freelist_new; - page->counters = counters_new; -- slab_unlock(page); -- if (disable_irqs) -- local_irq_restore(flags); -+ __slab_unlock(page, &flags, disable_irqs); - return true; - } -- slab_unlock(page); -- if (disable_irqs) -- local_irq_restore(flags); -+ __slab_unlock(page, &flags, disable_irqs); - } - - cpu_relax(); -@@ -1255,11 +1267,11 @@ static noinline int free_debug_processin - struct kmem_cache_node *n = get_node(s, page_to_nid(page)); - void *object = head; - int cnt = 0; -- unsigned long flags; -+ unsigned long flags, flags2; - int ret = 0; - - spin_lock_irqsave(&n->list_lock, flags); -- slab_lock(page); -+ slab_lock(page, &flags2); - - if (s->flags & SLAB_CONSISTENCY_CHECKS) { - if (!check_slab(s, page)) -@@ -1292,7 +1304,7 @@ static noinline int free_debug_processin - slab_err(s, page, "Bulk freelist count(%d) invalid(%d)\n", - bulk_cnt, cnt); - -- slab_unlock(page); -+ slab_unlock(page, &flags2); - spin_unlock_irqrestore(&n->list_lock, flags); - if (!ret) - slab_fix(s, "Object at 0x%p not freed", object); -@@ -4070,9 +4082,10 @@ static void list_slab_objects(struct kme - void *addr = page_address(page); - unsigned long *map; - void *p; -+ unsigned long flags; - - slab_err(s, page, text, s->name); -- slab_lock(page); -+ slab_lock(page, &flags); - - map = get_map(s, page); - for_each_object(p, s, addr, page->objects) { -@@ -4083,7 +4096,7 @@ static void list_slab_objects(struct kme - } - } - put_map(map); -- slab_unlock(page); -+ slab_unlock(page, &flags); - #endif - } - -@@ -4815,8 +4828,9 @@ static void validate_slab(struct kmem_ca - { - void *p; - void *addr = page_address(page); -+ unsigned long flags; - -- slab_lock(page); -+ slab_lock(page, &flags); - - if (!check_slab(s, page) || !on_freelist(s, page, NULL)) - goto unlock; -@@ -4831,7 +4845,7 @@ static void validate_slab(struct kmem_ca - break; - } - unlock: -- slab_unlock(page); -+ slab_unlock(page, &flags); - } - - static int validate_slab_node(struct kmem_cache *s, diff --git a/patches/0033-mm-slub-protect-put_cpu_partial-with-disabled-irqs-i.patch b/patches/0031-mm-slub-protect-put_cpu_partial-with-disabled-irqs-i.patch index 7ecc48a04726..84d123a5489d 100644 --- a/patches/0033-mm-slub-protect-put_cpu_partial-with-disabled-irqs-i.patch +++ b/patches/0031-mm-slub-protect-put_cpu_partial-with-disabled-irqs-i.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:30 +0200 -Subject: [PATCH 33/35] mm, slub: protect put_cpu_partial() with disabled irqs +Date: Wed, 28 Jul 2021 12:26:27 +0200 +Subject: [PATCH 31/33] mm, slub: protect put_cpu_partial() with disabled irqs instead of cmpxchg Jann Horn reported [1] the following theoretically possible race: @@ -44,14 +44,13 @@ reported by Jann should no longer be possible. Reported-by: Jann Horn <jannh@google.com> Suggested-by: Mike Galbraith <efault@gmx.de> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 83 +++++++++++++++++++++++++++++++++----------------------------- 1 file changed, 45 insertions(+), 38 deletions(-) --- a/mm/slub.c +++ b/mm/slub.c -@@ -2006,7 +2006,12 @@ static inline void *acquire_slab(struct +@@ -2025,7 +2025,12 @@ static inline void *acquire_slab(struct return freelist; } @@ -64,7 +63,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> static inline bool pfmemalloc_match(struct page *page, gfp_t gfpflags); /* -@@ -2440,14 +2445,6 @@ static void unfreeze_partials_cpu(struct +@@ -2459,14 +2464,6 @@ static void unfreeze_partials_cpu(struct __unfreeze_partials(s, partial_page); } @@ -79,7 +78,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * Put a page that was just frozen (in __slab_free|get_partial_node) into a * partial page slot if available. -@@ -2457,46 +2454,56 @@ static inline void unfreeze_partials_cpu +@@ -2476,46 +2473,56 @@ static inline void unfreeze_partials_cpu */ static void put_cpu_partial(struct kmem_cache *s, struct page *page, int drain) { @@ -163,6 +162,6 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> + +#endif /* CONFIG_SLUB_CPU_PARTIAL */ + - static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c, - bool lock) + static inline void flush_slab(struct kmem_cache *s, struct kmem_cache_cpu *c) { + unsigned long flags; diff --git a/patches/0032-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch b/patches/0032-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch deleted file mode 100644 index a87acd032ff6..000000000000 --- a/patches/0032-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch +++ /dev/null @@ -1,59 +0,0 @@ -From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:29 +0200 -Subject: [PATCH 32/35] mm, slub: make slab_lock() disable irqs with PREEMPT_RT - -We need to disable irqs around slab_lock() (a bit spinlock) to make it -irq-safe. The calls to slab_lock() are nested under spin_lock_irqsave() which -doesn't disable irqs on PREEMPT_RT, so add explicit disabling with PREEMPT_RT. - -We also distinguish cmpxchg_double_slab() where we do the disabling explicitly -and __cmpxchg_double_slab() for contexts with already disabled irqs. However -these context are also typically spin_lock_irqsave() thus insufficient on -PREEMPT_RT. Thus, change __cmpxchg_double_slab() to be same as -cmpxchg_double_slab() on PREEMPT_RT. - -Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> ---- - mm/slub.c | 13 +++++++++---- - 1 file changed, 9 insertions(+), 4 deletions(-) - ---- a/mm/slub.c -+++ b/mm/slub.c -@@ -380,12 +380,12 @@ static __always_inline void - static __always_inline void - slab_lock(struct page *page, unsigned long *flags) - { -- __slab_lock(page, flags, false); -+ __slab_lock(page, flags, IS_ENABLED(CONFIG_PREEMPT_RT)); - } - - static __always_inline void slab_unlock(struct page *page, unsigned long *flags) - { -- __slab_unlock(page, flags, false); -+ __slab_unlock(page, flags, IS_ENABLED(CONFIG_PREEMPT_RT)); - } - - static inline bool ___cmpxchg_double_slab(struct kmem_cache *s, struct page *page, -@@ -429,14 +429,19 @@ static inline bool ___cmpxchg_double_sla - return false; - } - --/* Interrupts must be disabled (for the fallback code to work right) */ -+/* -+ * Interrupts must be disabled (for the fallback code to work right), typically -+ * by an _irqsave() lock variant. Except on PREEMPT_RT where locks are different -+ * so we disable interrupts explicitly here. -+ */ - static inline bool __cmpxchg_double_slab(struct kmem_cache *s, struct page *page, - void *freelist_old, unsigned long counters_old, - void *freelist_new, unsigned long counters_new, - const char *n) - { - return ___cmpxchg_double_slab(s, page, freelist_old, counters_old, -- freelist_new, counters_new, n, false); -+ freelist_new, counters_new, n, -+ IS_ENABLED(CONFIG_PREEMPT_RT)); - } - - static inline bool cmpxchg_double_slab(struct kmem_cache *s, struct page *page, diff --git a/patches/0034-mm-slub-use-migrate_disable-on-PREEMPT_RT.patch b/patches/0032-mm-slub-use-migrate_disable-on-PREEMPT_RT.patch index 07772dcd3f4a..abeb6ac4d6a7 100644 --- a/patches/0034-mm-slub-use-migrate_disable-on-PREEMPT_RT.patch +++ b/patches/0032-mm-slub-use-migrate_disable-on-PREEMPT_RT.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:31 +0200 -Subject: [PATCH 34/35] mm, slub: use migrate_disable() on PREEMPT_RT +Date: Fri, 21 May 2021 14:03:23 +0200 +Subject: [PATCH 32/33] mm, slub: use migrate_disable() on PREEMPT_RT We currently use preempt_disable() (directly or via get_cpu_ptr()) to stabilize the pointer to kmem_cache_cpu. On PREEMPT_RT this would be incompatible with @@ -12,7 +12,6 @@ In order to get the best available mechanism on both PREEMPT_RT and wrappers and use them. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- mm/slub.c | 39 ++++++++++++++++++++++++++++++--------- 1 file changed, 30 insertions(+), 9 deletions(-) @@ -46,7 +45,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> #ifdef CONFIG_SLUB_DEBUG #ifdef CONFIG_SLUB_DEBUG_ON DEFINE_STATIC_KEY_TRUE(slub_debug_enabled); -@@ -2828,7 +2848,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2852,7 +2872,7 @@ static void *___slab_alloc(struct kmem_c if (unlikely(!pfmemalloc_match_unsafe(page, gfpflags))) goto deactivate_slab; @@ -55,7 +54,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> local_irq_save(flags); if (unlikely(page != c->page)) { local_irq_restore(flags); -@@ -2887,7 +2907,8 @@ static void *___slab_alloc(struct kmem_c +@@ -2911,7 +2931,8 @@ static void *___slab_alloc(struct kmem_c } if (unlikely(!slub_percpu_partial(c))) { local_irq_restore(flags); @@ -65,7 +64,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } page = c->page = slub_percpu_partial(c); -@@ -2903,9 +2924,9 @@ static void *___slab_alloc(struct kmem_c +@@ -2927,9 +2948,9 @@ static void *___slab_alloc(struct kmem_c if (freelist) goto check_new_page; @@ -77,7 +76,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (unlikely(!page)) { slab_out_of_memory(s, gfpflags, node); -@@ -2988,12 +3009,12 @@ static void *__slab_alloc(struct kmem_ca +@@ -3012,12 +3033,12 @@ static void *__slab_alloc(struct kmem_ca * cpu before disabling preemption. Need to reload cpu area * pointer. */ @@ -92,7 +91,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> #endif return p; } -@@ -3522,7 +3543,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3546,7 +3567,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca * IRQs, which protects against PREEMPT and interrupts * handlers invoking normal fastpath. */ @@ -101,7 +100,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> local_irq_disable(); for (i = 0; i < size; i++) { -@@ -3568,7 +3589,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3592,7 +3613,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca } c->tid = next_tid(c->tid); local_irq_enable(); @@ -110,7 +109,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * memcg and kmem_cache debug support and memory initialization. -@@ -3578,7 +3599,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3602,7 +3623,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca slab_want_init_on_alloc(flags, s)); return i; error: diff --git a/patches/0035-mm-slub-convert-kmem_cpu_slab-protection-to-local_lo.patch b/patches/0033-mm-slub-convert-kmem_cpu_slab-protection-to-local_lo.patch index 5778979fe3c7..0c04004739fc 100644 --- a/patches/0035-mm-slub-convert-kmem_cpu_slab-protection-to-local_lo.patch +++ b/patches/0033-mm-slub-convert-kmem_cpu_slab-protection-to-local_lo.patch @@ -1,6 +1,6 @@ From: Vlastimil Babka <vbabka@suse.cz> -Date: Thu, 29 Jul 2021 15:21:32 +0200 -Subject: [PATCH 35/35] mm, slub: convert kmem_cpu_slab protection to +Date: Sat, 22 May 2021 01:59:38 +0200 +Subject: [PATCH 33/33] mm, slub: convert kmem_cpu_slab protection to local_lock Embed local_lock into struct kmem_cpu_slab and use the irq-safe versions of @@ -27,11 +27,10 @@ by this series. [ Mike Galbraith <efault@gmx.de>: use local_lock() without irq in PREEMPT_RT scope; debugging of RT crashes resulting in put_cpu_partial() locking changes ] Signed-off-by: Vlastimil Babka <vbabka@suse.cz> -Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> --- include/linux/slub_def.h | 6 + - mm/slub.c | 142 ++++++++++++++++++++++++++++++++++++----------- - 2 files changed, 115 insertions(+), 33 deletions(-) + mm/slub.c | 146 +++++++++++++++++++++++++++++++++++------------ + 2 files changed, 117 insertions(+), 35 deletions(-) --- a/include/linux/slub_def.h +++ b/include/linux/slub_def.h @@ -147,7 +146,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> * * SLUB assigns one slab for allocation to each processor. * Allocations only occur from these slabs called cpu slabs. -@@ -2231,9 +2269,13 @@ static inline void note_cmpxchg_failure( +@@ -2250,9 +2288,13 @@ static inline void note_cmpxchg_failure( static void init_kmem_cache_cpus(struct kmem_cache *s) { int cpu; @@ -163,7 +162,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> } /* -@@ -2444,10 +2486,10 @@ static void unfreeze_partials(struct kme +@@ -2463,10 +2505,10 @@ static void unfreeze_partials(struct kme struct page *partial_page; unsigned long flags; @@ -176,7 +175,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (partial_page) __unfreeze_partials(s, partial_page); -@@ -2480,7 +2522,7 @@ static void put_cpu_partial(struct kmem_ +@@ -2499,7 +2541,7 @@ static void put_cpu_partial(struct kmem_ int pages = 0; int pobjects = 0; @@ -185,7 +184,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> oldpage = this_cpu_read(s->cpu_slab->partial); -@@ -2508,7 +2550,7 @@ static void put_cpu_partial(struct kmem_ +@@ -2527,7 +2569,7 @@ static void put_cpu_partial(struct kmem_ this_cpu_write(s->cpu_slab->partial, page); @@ -194,25 +193,43 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (page_to_unfreeze) { __unfreeze_partials(s, page_to_unfreeze); -@@ -2532,7 +2574,7 @@ static inline void flush_slab(struct kme +@@ -2549,7 +2591,7 @@ static inline void flush_slab(struct kme struct page *page; + void *freelist; - if (lock) -- local_irq_save(flags); -+ local_lock_irqsave(&s->cpu_slab->lock, flags); +- local_irq_save(flags); ++ local_lock_irqsave(&s->cpu_slab->lock, flags); - freelist = c->freelist; page = c->page; -@@ -2542,7 +2584,7 @@ static inline void flush_slab(struct kme + freelist = c->freelist; +@@ -2558,7 +2600,7 @@ static inline void flush_slab(struct kme + c->freelist = NULL; c->tid = next_tid(c->tid); - if (lock) -- local_irq_restore(flags); -+ local_unlock_irqrestore(&s->cpu_slab->lock, flags); +- local_irq_restore(flags); ++ local_unlock_irqrestore(&s->cpu_slab->lock, flags); - if (page) + if (page) { deactivate_slab(s, page, freelist); -@@ -2849,9 +2891,9 @@ static void *___slab_alloc(struct kmem_c +@@ -2780,8 +2822,6 @@ static inline bool pfmemalloc_match_unsa + * The page is still frozen if the return value is not NULL. + * + * If this function returns NULL then the page has been unfrozen. +- * +- * This function must be called with interrupt disabled. + */ + static inline void *get_freelist(struct kmem_cache *s, struct page *page) + { +@@ -2789,6 +2829,8 @@ static inline void *get_freelist(struct + unsigned long counters; + void *freelist; + ++ lockdep_assert_held(this_cpu_ptr(&s->cpu_slab->lock)); ++ + do { + freelist = page->freelist; + counters = page->counters; +@@ -2873,9 +2915,9 @@ static void *___slab_alloc(struct kmem_c goto deactivate_slab; /* must check again c->page in case we got preempted and it changed */ @@ -224,7 +241,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> goto reread_page; } freelist = c->freelist; -@@ -2862,7 +2904,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2886,7 +2928,7 @@ static void *___slab_alloc(struct kmem_c if (!freelist) { c->page = NULL; @@ -233,7 +250,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> stat(s, DEACTIVATE_BYPASS); goto new_slab; } -@@ -2871,7 +2913,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2895,7 +2937,7 @@ static void *___slab_alloc(struct kmem_c load_freelist: @@ -242,7 +259,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * freelist is pointing to the list of objects to be used. -@@ -2881,39 +2923,39 @@ static void *___slab_alloc(struct kmem_c +@@ -2905,39 +2947,39 @@ static void *___slab_alloc(struct kmem_c VM_BUG_ON(!c->page->frozen); c->freelist = get_freepointer(s, freelist); c->tid = next_tid(c->tid); @@ -290,7 +307,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> stat(s, CPU_PARTIAL_ALLOC); goto redo; } -@@ -2966,7 +3008,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2990,7 +3032,7 @@ static void *___slab_alloc(struct kmem_c retry_load_page: @@ -299,7 +316,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> if (unlikely(c->page)) { void *flush_freelist = c->freelist; struct page *flush_page = c->page; -@@ -2975,7 +3017,7 @@ static void *___slab_alloc(struct kmem_c +@@ -2999,7 +3041,7 @@ static void *___slab_alloc(struct kmem_c c->freelist = NULL; c->tid = next_tid(c->tid); @@ -308,7 +325,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> deactivate_slab(s, flush_page, flush_freelist); -@@ -3094,7 +3136,15 @@ static __always_inline void *slab_alloc_ +@@ -3118,7 +3160,15 @@ static __always_inline void *slab_alloc_ object = c->freelist; page = c->page; @@ -325,7 +342,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> object = __slab_alloc(s, gfpflags, node, addr, c); } else { void *next_object = get_freepointer_safe(s, object); -@@ -3354,6 +3404,7 @@ static __always_inline void do_slab_free +@@ -3378,6 +3428,7 @@ static __always_inline void do_slab_free barrier(); if (likely(page == c->page)) { @@ -333,7 +350,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> void **freelist = READ_ONCE(c->freelist); set_freepointer(s, tail_obj, freelist); -@@ -3366,6 +3417,31 @@ static __always_inline void do_slab_free +@@ -3390,6 +3441,31 @@ static __always_inline void do_slab_free note_cmpxchg_failure("slab_free", s, tid); goto redo; } @@ -365,7 +382,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> stat(s, FREE_FASTPATH); } else __slab_free(s, page, head, tail_obj, cnt, addr); -@@ -3544,7 +3620,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3568,7 +3644,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca * handlers invoking normal fastpath. */ c = slub_get_cpu_ptr(s->cpu_slab); @@ -374,7 +391,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> for (i = 0; i < size; i++) { void *object = kfence_alloc(s, s->object_size, flags); -@@ -3565,7 +3641,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3589,7 +3665,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca */ c->tid = next_tid(c->tid); @@ -383,7 +400,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> /* * Invoking slow path likely have side-effect -@@ -3579,7 +3655,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3603,7 +3679,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca c = this_cpu_ptr(s->cpu_slab); maybe_wipe_obj_freeptr(s, p[i]); @@ -392,7 +409,7 @@ Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> continue; /* goto for-loop */ } -@@ -3588,7 +3664,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca +@@ -3612,7 +3688,7 @@ int kmem_cache_alloc_bulk(struct kmem_ca maybe_wipe_obj_freeptr(s, p[i]); } c->tid = next_tid(c->tid); diff --git a/patches/Add_localversion_for_-RT_release.patch b/patches/Add_localversion_for_-RT_release.patch index 5da6b6218f1c..c0ab4419d1ec 100644 --- a/patches/Add_localversion_for_-RT_release.patch +++ b/patches/Add_localversion_for_-RT_release.patch @@ -15,4 +15,4 @@ Signed-off-by: Thomas Gleixner <tglx@linutronix.de> --- /dev/null +++ b/localversion-rt @@ -0,0 +1 @@ -+-rt18 ++-rt19 diff --git a/patches/locking-Remove-rt_rwlock_is_contended.patch b/patches/locking-Remove-rt_rwlock_is_contended.patch new file mode 100644 index 000000000000..06fb1d6c4216 --- /dev/null +++ b/patches/locking-Remove-rt_rwlock_is_contended.patch @@ -0,0 +1,33 @@ +From: Sebastian Andrzej Siewior <bigeasy@linutronix.de> +Date: Tue, 7 Sep 2021 12:11:47 +0200 +Subject: [PATCH] locking: Remove rt_rwlock_is_contended() + +rt_rwlock_is_contended() has not users. It makes no sense to use it as +rwlock_is_contended() because it is a sleeping lock on RT and preemption +is possible. It reports always != 0 if used by a writer and even if +there is a waiter then the lock might not be handed over if the +current owner has the highest priority. + +Remove rt_rwlock_is_contended(). + +Reported-by: kernel test robot <lkp@intel.com> +Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> +--- + kernel/locking/spinlock_rt.c | 6 ------ + 1 file changed, 6 deletions(-) + +--- a/kernel/locking/spinlock_rt.c ++++ b/kernel/locking/spinlock_rt.c +@@ -246,12 +246,6 @@ void __sched rt_write_unlock(rwlock_t *r + } + EXPORT_SYMBOL(rt_write_unlock); + +-int __sched rt_rwlock_is_contended(rwlock_t *rwlock) +-{ +- return rw_base_is_contended(&rwlock->rwbase); +-} +-EXPORT_SYMBOL(rt_rwlock_is_contended); +- + #ifdef CONFIG_DEBUG_LOCK_ALLOC + void __rt_rwlock_init(rwlock_t *rwlock, const char *name, + struct lock_class_key *key) diff --git a/patches/locking-rtmutex-Fix-ww_mutex-deadlock-check.patch b/patches/locking-rtmutex-Fix-ww_mutex-deadlock-check.patch new file mode 100644 index 000000000000..ead382f482cd --- /dev/null +++ b/patches/locking-rtmutex-Fix-ww_mutex-deadlock-check.patch @@ -0,0 +1,38 @@ +From: Peter Zijlstra <peterz@infradead.org> +Date: Wed, 1 Sep 2021 11:44:11 +0200 +Subject: [PATCH] locking/rtmutex: Fix ww_mutex deadlock check + +Dan reported that rt_mutex_adjust_prio_chain() can be called with +.orig_waiter == NULL however commit a055fcc132d4 ("locking/rtmutex: +Return success on deadlock for ww_mutex waiters") unconditionally +dereferences it. + +Since both call-sites that have .orig_waiter == NULL don't care for the +return value, simply disable the deadlock squash by adding the NULL +check. + +Notably, both callers use the deadlock condition as a termination +condition for the iteration; once detected, we're sure (de)boosting is +done. Arguably [3] would be a more natural termination point, but I'm +not sure adding a third deadlock detection state would improve the code. + +Fixes: a055fcc132d4 ("locking/rtmutex: Return success on deadlock for ww_mutex waiters") +Reported-by: Dan Carpenter <dan.carpenter@oracle.com> +Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org> +Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> +Link: https://lore.kernel.org/r/YS9La56fHMiCCo75@hirez.programming.kicks-ass.net +--- + kernel/locking/rtmutex.c | 2 +- + 1 file changed, 1 insertion(+), 1 deletion(-) + +--- a/kernel/locking/rtmutex.c ++++ b/kernel/locking/rtmutex.c +@@ -753,7 +753,7 @@ static int __sched rt_mutex_adjust_prio_ + * other configuration and we fail to report; also, see + * lockdep. + */ +- if (IS_ENABLED(CONFIG_PREEMPT_RT) && orig_waiter->ww_ctx) ++ if (IS_ENABLED(CONFIG_PREEMPT_RT) && orig_waiter && orig_waiter->ww_ctx) + ret = 0; + + raw_spin_unlock(&lock->wait_lock); diff --git a/patches/sched-Make-the-idle-timer-expire-always-in-hardirq-c.patch b/patches/sched-Make-the-idle-timer-expire-always-in-hardirq-c.patch new file mode 100644 index 000000000000..a10d1f9569d3 --- /dev/null +++ b/patches/sched-Make-the-idle-timer-expire-always-in-hardirq-c.patch @@ -0,0 +1,38 @@ +From: Sebastian Andrzej Siewior <bigeasy@linutronix.de> +Date: Mon, 6 Sep 2021 11:40:48 +0200 +Subject: [PATCH] sched: Make the idle timer expire always in hardirq context. + +The intel powerclamp driver will setup a per-CPU worker with RT +priority. The worker will then invoke play_idle() in which it remains in +the idle poll loop until it is stopped by the timer it started earlier. + +That timer needs to expire in hardirq context on PREEMPT_RT. Otherwise +the timer will expire in ksoftirqd as a SOFT timer but that task won't +be scheduled on the CPU because its priority is lower than the priority +of the worker which is in the idle loop. + +Always expire the idle timer in hardirq context. + +Fixes:c1de45ca831ac ("sched/idle: Add support for tasks that inject idle") +Reported-by: Thomas Gleixner <tglx@linutronix.de> +Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de> +Link: https://lkml.kernel.org/r/20210906113034.jgfxrjdvxnjqgtmc@linutronix.de +--- + kernel/sched/idle.c | 4 ++-- + 1 file changed, 2 insertions(+), 2 deletions(-) + +--- a/kernel/sched/idle.c ++++ b/kernel/sched/idle.c +@@ -379,10 +379,10 @@ void play_idle_precise(u64 duration_ns, + cpuidle_use_deepest_state(latency_ns); + + it.done = 0; +- hrtimer_init_on_stack(&it.timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL); ++ hrtimer_init_on_stack(&it.timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD); + it.timer.function = idle_inject_timer_fn; + hrtimer_start(&it.timer, ns_to_ktime(duration_ns), +- HRTIMER_MODE_REL_PINNED); ++ HRTIMER_MODE_REL_PINNED_HARD); + + while (!READ_ONCE(it.done)) + do_idle(); diff --git a/patches/series b/patches/series index 7fb2aa6ec52a..11ea9abed892 100644 --- a/patches/series +++ b/patches/series @@ -37,43 +37,41 @@ printk__Enhance_the_condition_check_of_msleep_in_pr_flush.patch ########################################################################### # mm bits polished by Mel and Vlastimil -# slub-local-lock-v4r3 +# slub-local-lock-v6r2 ########################################################################### 0001-mm-slub-don-t-call-flush_all-from-slab_debug_trace_o.patch 0002-mm-slub-allocate-private-object-map-for-debugfs-list.patch 0003-mm-slub-allocate-private-object-map-for-validate_sla.patch 0004-mm-slub-don-t-disable-irq-for-debug_check_no_locks_f.patch 0005-mm-slub-remove-redundant-unfreeze_partials-from-put_.patch -0006-mm-slub-unify-cmpxchg_double_slab-and-__cmpxchg_doub.patch -0007-mm-slub-extract-get_partial-from-new_slab_objects.patch -0008-mm-slub-dissolve-new_slab_objects-into-___slab_alloc.patch -0009-mm-slub-return-slab-page-from-get_partial-and-set-c-.patch -0010-mm-slub-restructure-new-page-checks-in-___slab_alloc.patch -0011-mm-slub-simplify-kmem_cache_cpu-and-tid-setup.patch -0012-mm-slub-move-disabling-enabling-irqs-to-___slab_allo.patch -0013-mm-slub-do-initial-checks-in-___slab_alloc-with-irqs.patch -0014-mm-slub-move-disabling-irqs-closer-to-get_partial-in.patch -0015-mm-slub-restore-irqs-around-calling-new_slab.patch -0016-mm-slub-validate-slab-from-partial-list-or-page-allo.patch -0017-mm-slub-check-new-pages-with-restored-irqs.patch -0018-mm-slub-stop-disabling-irqs-around-get_partial.patch -0019-mm-slub-move-reset-of-c-page-and-freelist-out-of-dea.patch -0020-mm-slub-make-locking-in-deactivate_slab-irq-safe.patch -0021-mm-slub-call-deactivate_slab-without-disabling-irqs.patch -0022-mm-slub-move-irq-control-into-unfreeze_partials.patch -0023-mm-slub-discard-slabs-in-unfreeze_partials-without-i.patch -0024-mm-slub-detach-whole-partial-list-at-once-in-unfreez.patch -0025-mm-slub-separate-detaching-of-partial-list-in-unfree.patch -0026-mm-slub-only-disable-irq-with-spin_lock-in-__unfreez.patch -0027-mm-slub-don-t-disable-irqs-in-slub_cpu_dead.patch -0028-mm-slab-make-flush_slab-possible-to-call-with-irqs-e.patch -0029-mm-slub-Move-flush_cpu_slab-invocations-__free_slab-.patch -0030-mm-slub-Make-object_map_lock-a-raw_spinlock_t.patch -0031-mm-slub-optionally-save-restore-irqs-in-slab_-un-loc.patch -0032-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch -0033-mm-slub-protect-put_cpu_partial-with-disabled-irqs-i.patch -0034-mm-slub-use-migrate_disable-on-PREEMPT_RT.patch -0035-mm-slub-convert-kmem_cpu_slab-protection-to-local_lo.patch +0006-mm-slub-extract-get_partial-from-new_slab_objects.patch +0007-mm-slub-dissolve-new_slab_objects-into-___slab_alloc.patch +0008-mm-slub-return-slab-page-from-get_partial-and-set-c-.patch +0009-mm-slub-restructure-new-page-checks-in-___slab_alloc.patch +0010-mm-slub-simplify-kmem_cache_cpu-and-tid-setup.patch +0011-mm-slub-move-disabling-enabling-irqs-to-___slab_allo.patch +0012-mm-slub-do-initial-checks-in-___slab_alloc-with-irqs.patch +0013-mm-slub-move-disabling-irqs-closer-to-get_partial-in.patch +0014-mm-slub-restore-irqs-around-calling-new_slab.patch +0015-mm-slub-validate-slab-from-partial-list-or-page-allo.patch +0016-mm-slub-check-new-pages-with-restored-irqs.patch +0017-mm-slub-stop-disabling-irqs-around-get_partial.patch +0018-mm-slub-move-reset-of-c-page-and-freelist-out-of-dea.patch +0019-mm-slub-make-locking-in-deactivate_slab-irq-safe.patch +0020-mm-slub-call-deactivate_slab-without-disabling-irqs.patch +0021-mm-slub-move-irq-control-into-unfreeze_partials.patch +0022-mm-slub-discard-slabs-in-unfreeze_partials-without-i.patch +0023-mm-slub-detach-whole-partial-list-at-once-in-unfreez.patch +0024-mm-slub-separate-detaching-of-partial-list-in-unfree.patch +0025-mm-slub-only-disable-irq-with-spin_lock-in-__unfreez.patch +0026-mm-slub-don-t-disable-irqs-in-slub_cpu_dead.patch +0027-mm-slab-split-out-the-cpu-offline-variant-of-flush_s.patch +0028-mm-slub-move-flush_cpu_slab-invocations-__free_slab-.patch +0029-mm-slub-make-object_map_lock-a-raw_spinlock_t.patch +0030-mm-slub-make-slab_lock-disable-irqs-with-PREEMPT_RT.patch +0031-mm-slub-protect-put_cpu_partial-with-disabled-irqs-i.patch +0032-mm-slub-use-migrate_disable-on-PREEMPT_RT.patch +0033-mm-slub-convert-kmem_cpu_slab-protection-to-local_lo.patch ########################################################################### # Posted @@ -82,6 +80,7 @@ highmem-Don-t-disable-preemption-on-RT-in-kmap_atomi.patch sched-Switch-wait_task_inactive-to-HRTIMER_MODE_REL_.patch sched-Prevent-balance_push-on-remote-runqueues.patch lockdep-Let-lock_is_held_type-detect-recursive-read-.patch +sched-Make-the-idle-timer-expire-always-in-hardirq-c.patch #KCOV 0001_documentation_kcov_include_types_h_in_the_example.patch @@ -227,6 +226,8 @@ locking-rtmutex-Prevent-spurious-EDEADLK-return-caus.patch ########################################################################### # Locking: RT bits. Need review ########################################################################### +locking-rtmutex-Fix-ww_mutex-deadlock-check.patch +locking-Remove-rt_rwlock_is_contended.patch lockdep-selftests-Avoid-using-local_lock_-acquire-re.patch 0001-sched-Trigger-warning-if-migration_disabled-counter-.patch 0003-rtmutex-Add-a-special-case-for-ww-mutex-handling.patch |