This new system call now provides all information required for libgtop.table_0_02

3 files changed, 440 insertions, 30 deletions

diff --git a/kernel/main.c b/kernel/main.c
index 7f6381bb..08a184d6 100644
--- a/kernel/main.c
+++ b/kernel/main.c
@@ -28,11 +28,168 @@
 
 #include "version.h"
 
+#if defined(__i386__)
+# define KSTK_EIP(tsk)	(((unsigned long *)tsk->kernel_stack_page)[1019])
+# define KSTK_ESP(tsk)	(((unsigned long *)tsk->kernel_stack_page)[1022])
+#elif defined(__alpha__)
+  /*
+   * See arch/alpha/kernel/ptrace.c for details.
+   */
+# define PT_REG(reg)		(PAGE_SIZE - sizeof(struct pt_regs)	\
+				 + (long)&((struct pt_regs *)0)->reg)
+# define KSTK_EIP(tsk)	(*(unsigned long *)(tsk->kernel_stack_page + PT_REG(pc)))
+# define KSTK_ESP(tsk)	((tsk) == current ? rdusp() : (tsk)->tss.usp)
+#elif defined(__sparc__)
+# define PT_REG(reg)            (PAGE_SIZE - sizeof(struct pt_regs)     \
+                                 + (long)&((struct pt_regs *)0)->reg)
+# define KSTK_EIP(tsk)  (*(unsigned long *)(tsk->kernel_stack_page + PT_REG(pc)))
+# define KSTK_ESP(tsk)  (*(unsigned long *)(tsk->kernel_stack_page + PT_REG(u_regs[UREG_FP])))
+#endif
+
+static struct task_struct *
+get_task (pid_t pid)
+{
+	struct task_struct ** p;
+	
+	p = task;
+	while (++p < task+NR_TASKS) {
+		if (*p && (*p)->pid == pid)
+			return *p;
+	}
+
+	return NULL;
+}
+
+static inline void statm_pte_range(pmd_t * pmd, unsigned long address, unsigned long size,
+	int * pages, int * shared, int * dirty, int * total)
+{
+	pte_t * pte;
+	unsigned long end;
+
+	if (pmd_none(*pmd))
+		return;
+	if (pmd_bad(*pmd)) {
+		printk("statm_pte_range: bad pmd (%08lx)\n", pmd_val(*pmd));
+		pmd_clear(pmd);
+		return;
+	}
+	pte = pte_offset(pmd, address);
+	address &= ~PMD_MASK;
+	end = address + size;
+	if (end > PMD_SIZE)
+		end = PMD_SIZE;
+	do {
+		pte_t page = *pte;
+
+		address += PAGE_SIZE;
+		pte++;
+		if (pte_none(page))
+			continue;
+		++*total;
+		if (!pte_present(page))
+			continue;
+		++*pages;
+		if (pte_dirty(page))
+			++*dirty;
+		if (pte_page(page) >= high_memory)
+			continue;
+		if (mem_map[MAP_NR(pte_page(page))].count > 1)
+			++*shared;
+	} while (address < end);
+}
+
+static inline void statm_pmd_range(pgd_t * pgd, unsigned long address, unsigned long size,
+	int * pages, int * shared, int * dirty, int * total)
+{
+	pmd_t * pmd;
+	unsigned long end;
+
+	if (pgd_none(*pgd))
+		return;
+	if (pgd_bad(*pgd)) {
+		printk("statm_pmd_range: bad pgd (%08lx)\n", pgd_val(*pgd));
+		pgd_clear(pgd);
+		return;
+	}
+	pmd = pmd_offset(pgd, address);
+	address &= ~PGDIR_MASK;
+	end = address + size;
+	if (end > PGDIR_SIZE)
+		end = PGDIR_SIZE;
+	do {
+		statm_pte_range(pmd, address, end - address, pages, shared, dirty, total);
+		address = (address + PMD_SIZE) & PMD_MASK;
+		pmd++;
+	} while (address < end);
+}
+
+static void statm_pgd_range(pgd_t * pgd, unsigned long address, unsigned long end,
+	int * pages, int * shared, int * dirty, int * total)
+{
+	while (address < end) {
+		statm_pmd_range(pgd, address, end - address, pages, shared, dirty, total);
+		address = (address + PGDIR_SIZE) & PGDIR_MASK;
+		pgd++;
+	}
+}
+
+static unsigned long
+get_wchan (struct task_struct *p)
+{
+	if (!p || p == current || p->state == TASK_RUNNING)
+		return 0;
+#if defined(__i386__)
+	{
+		unsigned long ebp, eip;
+		unsigned long stack_page;
+		int count = 0;
+
+		stack_page = p->kernel_stack_page;
+		if (!stack_page)
+			return 0;
+		ebp = p->tss.ebp;
+		do {
+			if (ebp < stack_page || ebp >= 4092+stack_page)
+				return 0;
+			eip = *(unsigned long *) (ebp+4);
+			if (eip < (unsigned long) interruptible_sleep_on
+			    || eip >= (unsigned long) add_timer)
+				return eip;
+			ebp = *(unsigned long *) ebp;
+		} while (count++ < 16);
+	}
+#elif defined(__alpha__)
+	/*
+	 * This one depends on the frame size of schedule().  Do a
+	 * "disass schedule" in gdb to find the frame size.  Also, the
+	 * code assumes that sleep_on() follows immediately after
+	 * interruptible_sleep_on() and that add_timer() follows
+	 * immediately after interruptible_sleep().  Ugly, isn't it?
+	 * Maybe adding a wchan field to task_struct would be better,
+	 * after all...
+	 */
+	{
+		unsigned long schedule_frame;
+		unsigned long pc;
+		
+		pc = thread_saved_pc(&p->tss);
+		if (pc >= (unsigned long) interruptible_sleep_on && pc < (unsigned long) add_timer) {
+			schedule_frame = ((unsigned long *)p->tss.ksp)[6];
+			return ((unsigned long *)schedule_frame)[12];
+		}
+		return pc;
+	}
+#endif
+	return 0;
+}
+
 asmlinkage int
-sys_table (int type, union table *buf)
+sys_table (int type, union table *buf, const void *param)
 {
 	union table tbl;
 	struct sysinfo i;
+	struct task_struct *tsk = NULL;
+	pid_t pid;
 	int err;
 
 	if (type == TABLE_VERSION)
@@ -41,6 +198,31 @@ sys_table (int type, union table *buf)
 	if (!buf)
 		return -EFAULT;
 
+	memset (&tbl, 0, sizeof (union table));
+
+	/* For TABLE_PROC_*, read pid and get task_struct */
+
+	switch (type) {
+	case TABLE_PROC_UID:
+	case TABLE_PROC_MEM:
+	case TABLE_PROC_SEGMENT:
+	case TABLE_PROC_TIME:
+	case TABLE_PROC_STATE:
+	case TABLE_PROC_SIGNAL:
+	case TABLE_PROC_KERNEL:
+	  err = verify_area (VERIFY_READ, param, sizeof (pid_t));
+		if (err)
+			return err;
+		memcpy_fromfs (&pid, param, sizeof (pid_t));
+
+		tsk = get_task (pid);
+		if (tsk == NULL)
+			return -ESRCH;
+		break;
+	}
+
+	/* Main function dispatcher */
+
 	switch (type) {
 	case TABLE_CPU:
 		tbl.cpu.total = jiffies;    
@@ -66,9 +248,9 @@ sys_table (int type, union table *buf)
 		tbl.swap.free  = i.freeswap;
 		break;
 	case TABLE_LOADAVG:
-		tbl.loadavg.loadavg [0] = avenrun [0];
-		tbl.loadavg.loadavg [1] = avenrun [1];
-		tbl.loadavg.loadavg [2] = avenrun [2];
+		tbl.loadavg.loadavg [0] = (double) avenrun [0] / (1 << FSHIFT);
+		tbl.loadavg.loadavg [1] = (double) avenrun [1] / (1 << FSHIFT);
+		tbl.loadavg.loadavg [2] = (double) avenrun [2] / (1 << FSHIFT);
 		tbl.loadavg.nr_running = nr_running;
 		tbl.loadavg.nr_tasks = nr_tasks;
 		tbl.loadavg.last_pid = last_pid;
@@ -77,6 +259,157 @@ sys_table (int type, union table *buf)
 		tbl.uptime.uptime = jiffies;
 		tbl.uptime.idle   = task[0]->utime + task[0]->stime;
 		break;
+	case TABLE_PROC_STATE:
+		tbl.proc_state.state = tsk->state;
+		tbl.proc_state.flags = tsk->flags;
+		memcpy (tbl.proc_state.comm, tsk->comm, sizeof (tbl.proc_state.comm));
+		break;
+	case TABLE_PROC_UID:
+		tbl.proc_uid.uid = tsk->uid;
+		tbl.proc_uid.euid = tsk->euid;
+		tbl.proc_uid.suid = tsk->suid;
+		tbl.proc_uid.fsuid = tsk->fsuid;
+		
+		tbl.proc_uid.gid = tsk->gid;
+		tbl.proc_uid.egid = tsk->egid;
+		tbl.proc_uid.sgid = tsk->sgid;
+		tbl.proc_uid.fsgid = tsk->fsgid;
+		
+		tbl.proc_uid.pid = tsk->pid;
+		tbl.proc_uid.pgrp = tsk->pgrp;
+		tbl.proc_uid.ppid = tsk->p_pptr->pid;
+		
+		tbl.proc_uid.session = tsk->session;
+	        tbl.proc_uid.tty = tsk->tty ? kdev_t_to_nr (tsk->tty->device) : 0;
+		tbl.proc_uid.tpgid = tsk->tty ? tsk->tty->pgrp : -1;
+		
+		tbl.proc_uid.priority = tsk->priority;
+		tbl.proc_uid.counter = tsk->counter;
+		tbl.proc_uid.def_priority = DEF_PRIORITY;
+		break;
+	case TABLE_PROC_SIGNAL:
+		tbl.proc_signal.signal = tsk->signal;
+		tbl.proc_signal.blocked = tsk->blocked;
+
+		if (tsk->sig) {
+			struct sigaction * action = tsk->sig->action;
+			unsigned long sig_ign = 0, sig_caught = 0;
+			unsigned long bit = 1;
+			int i;
+			
+			for (i = 0; i < 32; i++) {
+				switch((unsigned long) action->sa_handler) {
+				case 0:
+					break;
+				case 1:
+					sig_ign |= bit;
+					break;
+				default:
+					sig_caught |= bit;
+				}
+				bit <<= 1;
+				action++;
+			}
+
+			tbl.proc_signal.ignored = sig_ign;
+			tbl.proc_signal.caught  = sig_caught;
+		} else {
+			tbl.proc_signal.ignored = 0;
+			tbl.proc_signal.caught  = 0;
+		}
+		break;
+	case TABLE_PROC_MEM:
+		if (tsk->mm && tsk->mm != &init_mm) {
+			tbl.proc_mem.context = tsk->mm->context;
+			tbl.proc_mem.start_code = tsk->mm->start_code;
+			tbl.proc_mem.end_code = tsk->mm->end_code;
+			tbl.proc_mem.start_data = tsk->mm-> start_data;
+			tbl.proc_mem.end_data = tsk->mm->end_data;
+			tbl.proc_mem.start_brk = tsk->mm->start_brk;
+			tbl.proc_mem.brk = tsk->mm->brk;
+			tbl.proc_mem.start_stack = tsk->mm->start_stack;
+			tbl.proc_mem.start_mmap = tsk->mm->start_mmap;
+			tbl.proc_mem.arg_start = tsk->mm->arg_start;
+			tbl.proc_mem.arg_end = tsk->mm->arg_end;
+			tbl.proc_mem.env_start = tsk->mm->env_start;
+			tbl.proc_mem.env_end = tsk->mm->env_end;
+			tbl.proc_mem.rss = tsk->mm->rss;
+			tbl.proc_mem.total_vm = tsk->mm->total_vm;
+			tbl.proc_mem.locked_vm = tsk->mm->locked_vm;
+		}
+		tbl.proc_mem.rlim = tsk->rlim ? tsk->rlim[RLIMIT_RSS].rlim_cur : 0;
+		break;
+	case TABLE_PROC_SEGMENT:
+		if (tsk->mm && tsk->mm != &init_mm) {
+			unsigned long vsize = 0;
+			int size = 0, resident = 0, share = 0;
+			int trs = 0, lrs = 0, drs = 0, dt = 0;
+			struct vm_area_struct * vma = tsk->mm->mmap;
+
+			while (vma) {
+				pgd_t *pgd = pgd_offset(tsk->mm, vma->vm_start);
+				int pages = 0, shared = 0, dirty = 0, total = 0;
+				
+				vsize += vma->vm_end - vma->vm_start;
+
+				statm_pgd_range (pgd, vma->vm_start, vma->vm_end,
+						 &pages, &shared, &dirty, &total);
+				resident += pages;
+				share += shared;
+				dt += dirty;
+				size += total;
+				if (vma->vm_flags & VM_EXECUTABLE)
+					trs += pages;	/* text */
+				else if (vma->vm_flags & VM_GROWSDOWN)
+					drs += pages;	/* stack */
+				else if (vma->vm_end > 0x60000000)
+					lrs += pages;	/* library */
+				else
+					drs += pages;
+				vma = vma->vm_next;
+			}
+
+			tbl.proc_segment.vsize = vsize;
+			tbl.proc_segment.size = size;
+			tbl.proc_segment.resident = resident;
+			tbl.proc_segment.shared = share;
+			tbl.proc_segment.trs = trs;
+			tbl.proc_segment.lrs = lrs;
+			tbl.proc_segment.dt = dt;
+		}
+		break;
+	case TABLE_PROC_TIME:
+		tbl.proc_time.utime = tsk->utime;
+		tbl.proc_time.stime = tsk->stime;
+		tbl.proc_time.cutime = tsk->cutime;
+		tbl.proc_time.cstime = tsk->cstime;
+
+		tbl.proc_time.start_time = tsk->start_time;
+		tbl.proc_time.timeout = tsk->timeout;
+		tbl.proc_time.policy = tsk->policy;
+		tbl.proc_time.rt_priority = tsk->rt_priority;
+
+		tbl.proc_time.it_real_value = tsk->it_real_value;
+		tbl.proc_time.it_prof_value = tsk->it_prof_value;
+		tbl.proc_time.it_virt_value = tsk->it_virt_value;
+		tbl.proc_time.it_real_incr = tsk->it_real_incr;
+		tbl.proc_time.it_prof_incr = tsk->it_prof_incr;
+		tbl.proc_time.it_virt_incr = tsk->it_virt_incr;
+		break;
+	case TABLE_PROC_KERNEL:
+		tbl.proc_kernel.min_flt = tsk->min_flt;
+		tbl.proc_kernel.cmin_flt = tsk->cmin_flt;
+		tbl.proc_kernel.maj_flt = tsk->maj_flt;
+		tbl.proc_kernel.cmaj_flt = tsk->cmaj_flt;
+		
+		tbl.proc_kernel.kesp = tsk->kernel_stack_page ? KSTK_EIP(tsk) : 0;
+		tbl.proc_kernel.keip = tsk->kernel_stack_page ? KSTK_ESP(tsk) : 0;
+		
+		tbl.proc_kernel.nswap = tsk->nswap;
+		tbl.proc_kernel.cnswap = tsk->cnswap;
+
+		tbl.proc_kernel.wchan = get_wchan (tsk);
+		break;
 	default:
 		return -EINVAL;
 	}
diff --git a/kernel/table.h b/kernel/table.h
index 9515fdc9..801b410b 100644
--- a/kernel/table.h
+++ b/kernel/table.h
@@ -1,13 +1,23 @@
 #ifndef _LINUX_TABLE_H
 #define _LINUX_TABLE_H
-#include <linux/types.h>
 
-#define TABLE_VERSION	0
-#define TABLE_CPU	1
-#define TABLE_MEM	2
-#define TABLE_SWAP	3
-#define TABLE_LOADAVG	4
-#define TABLE_UPTIME	5
+#ifdef _KERNEL
+#include <linux/types.h>
+#endif
+
+#define TABLE_VERSION		0
+#define TABLE_CPU		1
+#define TABLE_MEM		2
+#define TABLE_SWAP		3
+#define TABLE_LOADAVG		4
+#define TABLE_UPTIME		5
+#define TABLE_PROC_UID		6
+#define TABLE_PROC_MEM		7
+#define TABLE_PROC_SEGMENT	8
+#define TABLE_PROC_TIME		9
+#define TABLE_PROC_STATE	10
+#define TABLE_PROC_SIGNAL	11
+#define TABLE_PROC_KERNEL	12
 
 /* CPU Usage (in jiffies = 1/100th seconds) */
 
@@ -46,7 +56,7 @@ struct table_swap
 
 struct table_loadavg
 {
-	unsigned long loadavg [3];
+	double loadavg [3];
 	unsigned nr_running;
 	unsigned nr_tasks;
 	unsigned last_pid;
@@ -60,6 +70,67 @@ struct table_uptime
 	unsigned long idle;
 };
 
+/* Information about processes. */
+
+struct table_proc_state
+{
+	long state;
+	unsigned long flags;
+	char comm[16];
+};
+
+struct table_proc_uid
+{
+	int uid, euid, suid, fsuid;
+	int gid, egid, sgid, fsgid;
+	int pid, pgrp, ppid;
+	int session;
+	unsigned int tty;
+	int tpgid;
+	long priority;
+	long counter;
+	long def_priority;
+};
+
+struct table_proc_mem
+{
+	unsigned long context;
+	unsigned long start_code, end_code, start_data, end_data;
+	unsigned long start_brk, brk, start_stack, start_mmap;
+	unsigned long arg_start, arg_end, env_start, env_end;
+	unsigned long rss, rlim, total_vm, locked_vm;
+};
+
+struct table_proc_segment
+{
+	unsigned long vsize;
+	int size, resident, shared;
+	int trs, lrs, drs, dt;
+};
+
+struct table_proc_time
+{
+	long utime, stime, cutime, cstime, start_time;
+	unsigned long timeout, policy, rt_priority;
+	unsigned long it_real_value, it_prof_value, it_virt_value;
+	unsigned long it_real_incr, it_prof_incr, it_virt_incr;
+};
+
+struct table_proc_signal
+{
+	unsigned long signal;
+	unsigned long blocked;	/* bitmap of masked signals */
+	unsigned long ignored;  /* mask of ignored signals */
+	unsigned long caught;   /* mask of caught signals */
+};
+
+struct table_proc_kernel
+{
+	unsigned long keip, kesp, wchan;
+	unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
+	unsigned long nswap, cnswap;
+};
+
 /* Union */
 
 union table
@@ -69,12 +140,15 @@ union table
 	struct table_swap swap;
 	struct table_loadavg loadavg;
 	struct table_uptime uptime;
+	struct table_proc_uid proc_uid;
+	struct table_proc_mem proc_mem;
+	struct table_proc_segment proc_segment;
+	struct table_proc_time proc_time;
+	struct table_proc_state proc_state;	
+	struct table_proc_signal proc_signal;
+	struct table_proc_kernel proc_kernel;
 };
 
-#ifdef __KERNEL__
-
-#endif /* __KERNEL__ */
-
 #endif /* _LINUX_IPC_H */
 
 
diff --git a/kernel/test.c b/kernel/test.c
index a6ba8dc7..f4ed092c 100644
--- a/kernel/test.c
+++ b/kernel/test.c
@@ -7,15 +7,16 @@
 
 #include <syscall.h>
 
-static inline _syscall2 (int, table, int, type, union table *, tbl);
+static inline _syscall3 (int, table, int, type, union table *, tbl, const void *, param);
 
 int
 main (void)
 {
 	union table tbl;
+	unsigned count;
 	int ret;
 
-	ret = table (TABLE_VERSION, NULL);
+	ret = table (TABLE_VERSION, NULL, NULL);
 
 	if (ret == -1) {
 		fprintf (stderr, "table(%u): %s\n", TABLE_VERSION, sys_errlist [errno]);
@@ -24,18 +25,20 @@ main (void)
 
 	fprintf (stderr, "Table (%u) = %u\n", TABLE_VERSION, ret);
 
-	ret = table (TABLE_CPU, &tbl);
+	for (count = 0; count < 5; count++) {
+		ret = table (TABLE_CPU, &tbl, NULL);
 
-	if (ret == -1) {
-		fprintf (stderr, "table(%u): %s\n", TABLE_CPU, sys_errlist [errno]);
-		exit (-errno);
-	}
+		if (ret == -1) {
+			fprintf (stderr, "table(%u): %s\n", TABLE_CPU, sys_errlist [errno]);
+			exit (-errno);
+		}
 
-	fprintf (stderr, "Table (%u) = %lu, %lu, %lu, %lu, %lu, %lu\n",
-		 TABLE_CPU, tbl.cpu.total, tbl.cpu.user, tbl.cpu.nice,
-		 tbl.cpu.sys, tbl.cpu.idle, tbl.cpu.frequency);
+		fprintf (stderr, "Table (%u) = %lu, %lu, %lu, %lu, %lu, %lu\n",
+			 TABLE_CPU, tbl.cpu.total, tbl.cpu.user, tbl.cpu.nice,
+			 tbl.cpu.sys, tbl.cpu.idle, tbl.cpu.frequency);
+	}
 
-	ret = table (TABLE_MEM, &tbl);
+	ret = table (TABLE_MEM, &tbl, NULL);
 
 	if (ret == -1) {
 		fprintf (stderr, "table(%u): %s\n", TABLE_MEM, sys_errlist [errno]);
@@ -46,7 +49,7 @@ main (void)
 		 TABLE_MEM, tbl.mem.total, tbl.mem.used, tbl.mem.free,
 		 tbl.mem.shared, tbl.mem.buffer, tbl.mem.cached);
 
-	ret = table (TABLE_SWAP, &tbl);
+	ret = table (TABLE_SWAP, &tbl, NULL);
 
 	if (ret == -1) {
 		fprintf (stderr, "table(%u): %s\n", TABLE_SWAP, sys_errlist [errno]);
@@ -56,7 +59,7 @@ main (void)
 	fprintf (stderr, "Table (%u) = %lu, %lu, %lu\n",
 		 TABLE_SWAP, tbl.swap.total, tbl.swap.used, tbl.swap.free);
 	
-	ret = table (TABLE_LOADAVG, &tbl);
+	ret = table (TABLE_LOADAVG, &tbl, NULL);
 
 	if (ret == -1) {
 		fprintf (stderr, "table(%u): %s\n", TABLE_LOADAVG, sys_errlist [errno]);
@@ -68,7 +71,7 @@ main (void)
 		 tbl.loadavg.loadavg [2], tbl.loadavg.nr_running,
 		 tbl.loadavg.nr_tasks, tbl.loadavg.last_pid);
 	
-	ret = table (TABLE_UPTIME, &tbl);
+	ret = table (TABLE_UPTIME, &tbl, NULL);
 
 	if (ret == -1) {
 		fprintf (stderr, "table(%u): %s\n", TABLE_UPTIME, sys_errlist [errno]);