diff options
author | Simon Marlow <marlowsd@gmail.com> | 2011-10-27 13:47:27 +0100 |
---|---|---|
committer | Simon Marlow <marlowsd@gmail.com> | 2011-11-02 16:34:05 +0000 |
commit | 7bb0447df9a783c222c2a077e35e5013c7c68d91 (patch) | |
tree | 78d6d2a14f7e42df5cda32199c71ced973f169ef /utils/genapply | |
parent | bd72eeb184a95ae0ae79ccad19c8ccc2b45a12e0 (diff) | |
download | haskell-7bb0447df9a783c222c2a077e35e5013c7c68d91.tar.gz |
Overhaul of infrastructure for profiling, coverage (HPC) and breakpoints
User visible changes
====================
Profilng
--------
Flags renamed (the old ones are still accepted for now):
OLD NEW
--------- ------------
-auto-all -fprof-auto
-auto -fprof-exported
-caf-all -fprof-cafs
New flags:
-fprof-auto Annotates all bindings (not just top-level
ones) with SCCs
-fprof-top Annotates just top-level bindings with SCCs
-fprof-exported Annotates just exported bindings with SCCs
-fprof-no-count-entries Do not maintain entry counts when profiling
(can make profiled code go faster; useful with
heap profiling where entry counts are not used)
Cost-centre stacks have a new semantics, which should in most cases
result in more useful and intuitive profiles. If you find this not to
be the case, please let me know. This is the area where I have been
experimenting most, and the current solution is probably not the
final version, however it does address all the outstanding bugs and
seems to be better than GHC 7.2.
Stack traces
------------
+RTS -xc now gives more information. If the exception originates from
a CAF (as is common, because GHC tends to lift exceptions out to the
top-level), then the RTS walks up the stack and reports the stack in
the enclosing update frame(s).
Result: +RTS -xc is much more useful now - but you still have to
compile for profiling to get it. I've played around a little with
adding 'head []' to GHC itself, and +RTS -xc does pinpoint the problem
quite accurately.
I plan to add more facilities for stack tracing (e.g. in GHCi) in the
future.
Coverage (HPC)
--------------
* derived instances are now coloured yellow if they weren't used
* likewise record field names
* entry counts are more accurate (hpc --fun-entry-count)
* tab width is now correct (markup was previously off in source with
tabs)
Internal changes
================
In Core, the Note constructor has been replaced by
Tick (Tickish b) (Expr b)
which is used to represent all the kinds of source annotation we
support: profiling SCCs, HPC ticks, and GHCi breakpoints.
Depending on the properties of the Tickish, different transformations
apply to Tick. See CoreUtils.mkTick for details.
Tickets
=======
This commit closes the following tickets, test cases to follow:
- Close #2552: not a bug, but the behaviour is now more intuitive
(test is T2552)
- Close #680 (test is T680)
- Close #1531 (test is result001)
- Close #949 (test is T949)
- Close #2466: test case has bitrotted (doesn't compile against current
version of vector-space package)
Diffstat (limited to 'utils/genapply')
-rw-r--r-- | utils/genapply/GenApply.hs | 65 |
1 files changed, 51 insertions, 14 deletions
diff --git a/utils/genapply/GenApply.hs b/utils/genapply/GenApply.hs index d9e6041a61..2ffa81bb76 100644 --- a/utils/genapply/GenApply.hs +++ b/utils/genapply/GenApply.hs @@ -230,8 +230,10 @@ genMkPAP regstatus macro jump ticker disamb else empty, if is_fun_case then mb_tag_node arity else empty, - text "jump " <> text jump <> semi - ]) $$ + if overflow_regs + then text "jump_SAVE_CCCS" <> parens (text jump) <> semi + else text "jump " <> text jump <> semi + ]) $$ text "}" where @@ -280,18 +282,37 @@ genMkPAP regstatus macro jump ticker disamb loadSpWordOff "W_" 0 <> text " = " <> mkApplyInfoName rest_args <> semi - shuffle_extra_args - = vcat (map shuffle_down - [sp_stk_args .. sp_stk_args+stack_args_size-1]) $$ - loadSpWordOff "W_" (sp_stk_args+stack_args_size-1) - <> text " = " - <> mkApplyInfoName rest_args <> semi $$ - text "Sp_adj(" <> int (sp_stk_args - 1) <> text ");" - - shuffle_down i = - loadSpWordOff "W_" (i-1) <> text " = " <> + shuffle_extra_args + = vcat [text "#ifdef PROFILING", + shuffle True, + text "#else", + shuffle False, + text "#endif"] + where + -- Sadly here we have to insert an stg_restore_cccs frame + -- just underneath the stg_ap_*_info frame if we're + -- profiling; see Note [jump_SAVE_CCCS] + shuffle prof = + let offset = if prof then 2 else 0 in + vcat (map (shuffle_down (offset+1)) + [sp_stk_args .. sp_stk_args+stack_args_size-1]) $$ + (if prof + then + loadSpWordOff "W_" (sp_stk_args+stack_args_size-3) + <> text " = stg_restore_cccs_info;" $$ + loadSpWordOff "W_" (sp_stk_args+stack_args_size-2) + <> text " = W_[CCCS];" + else empty) $$ + loadSpWordOff "W_" (sp_stk_args+stack_args_size-1) + <> text " = " + <> mkApplyInfoName rest_args <> semi $$ + text "Sp_adj(" <> int (sp_stk_args - 1 - offset) <> text ");" + + shuffle_down j i = + loadSpWordOff "W_" (i-j) <> text " = " <> loadSpWordOff "W_" i <> semi + -- The EXACT ARITY case -- -- if (arity == 1) { @@ -357,6 +378,21 @@ genMkPAP regstatus macro jump ticker disamb = assignRegs regstatus stk_args_slow_offset args -- BUILD_PAP assumes args start at offset 1 +-- Note [jump_SAVE_CCCS] + +-- when profiling, if we have some extra arguments to apply that we +-- save to the stack, we must also save the current cost centre stack +-- and restore it when applying the extra arguments. This is all +-- handled by the macro jump_SAVE_CCCS(target), defined in +-- rts/AutoApply.h. +-- +-- At the jump, the stack will look like this: +-- +-- ... extra args ... +-- stg_ap_pp_info +-- CCCS +-- stg_restore_cccs_info + -- -------------------------------------- -- Examine tag bits of function pointer and enter it -- directly if needed. @@ -579,8 +615,9 @@ genApply regstatus args = -- overwritten by an indirection, so we must enter the original -- info pointer we read, don't read it again, because it might -- not be enterable any more. - text "jump %ENTRY_CODE(info);", - text "" + text "jump_SAVE_CCCS(%ENTRY_CODE(info));", + -- see Note [jump_SAVE_CCCS] + text "" ]), text "}", |