Linux 2.6.24.4+mmap with overcommit >= 1

Linux 2.6.24.4+mmap with overcommit >= 1

[Sami Farin]

[-- Attachment #1: Type: text/plain, Size: 4359 bytes --]

I have arch x86_64, glibc-2.7.90-7 from Fedora.
When I have vm.overcommit_memory == 1 or 2,
mmap succeeds for really huge values.
These ranges succeed (approximate):
 8793870000000 - 17500387000000
26390387000000 - 35180387000000

with overcommit_memory=0 it behaves predictably.

03:49:57.378780 write(2, "trying 27386666640000 bytes...", 30trying 27386666640000 bytes...) = 30 <0.000014>
03:49:57.378914 mmap(NULL, 27386666643456, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x2ada07c88000 <0.000013>
03:49:57.378976 write(2, " got 0x2ada07c88010\n", 20 got 0x2ada07c88010
) = 20 <0.000014>
03:49:57.804125 --- SIGINT (Interrupt) @ 0 (0) ---

but with ten times smaller value:
03:49:40.051589 write(2, "trying 2738666664000 bytes...", 29trying 2738666664000 bytes...) = 29 <0.000014>
03:49:40.051722 mmap(NULL, 2738666668032, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = -1 ENOMEM (Cannot allocate memory) <0.000012>
03:49:40.051783 brk(0)                  = 0x1362000 <0.000011>
03:49:40.051830 brk(0x27da6792000)      = 0x1362000 <0.000012>
03:49:40.051881 mmap(NULL, 2738666799104, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = -1 ENOMEM (Cannot allocate memory) <0.000013>
03:49:40.051942 mmap(NULL, 134217728, PROT_NONE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_NORESERVE, -1, 0) = 0x3053f4e47000 <0.000013>
03:49:40.051993 munmap(0x3053f4e47000, 52137984) = 0 <0.000016>
03:49:40.052045 munmap(0x3053fc000000, 14970880) = 0 <0.000015>
03:49:40.052095 mprotect(0x3053f8000000, 135168, PROT_READ|PROT_WRITE) = 0 <0.000014>
03:49:40.052161 mmap(NULL, 2738666668032, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = -1 ENOMEM (Cannot allocate memory) <0.000013>
03:49:40.052220 write(2, " got (nil)\n", 11 got (nil)

MemTotal:      3062112 kB
MemFree:         93900 kB
Buffers:             0 kB
Cached:         104652 kB
SwapCached:     176576 kB
Active:        1504444 kB
Inactive:       434308 kB
SwapTotal:     3911784 kB
SwapFree:      2947756 kB
Dirty:             288 kB
Writeback:           0 kB
AnonPages:     1815468 kB
Mapped:          92448 kB
Slab:           836956 kB
SReclaimable:   647004 kB
SUnreclaim:     189952 kB
PageTables:      24096 kB
NFS_Unstable:        0 kB
Bounce:              0 kB
CommitLimit:   5442840 kB
Committed_AS: 18446744066096782940 kB
VmallocTotal: 34359738367 kB
VmallocUsed:     11940 kB
VmallocChunk: 34359725791 kB
HugePages_Total:     0
HugePages_Free:      0
HugePages_Rsvd:      0
HugePages_Surp:      0
Hugepagesize:     2048 kB

[2622805.705953] wmweather+[6197]: segfault at 28 rip 555555565713 rsp 7fff2ba47e10 error 4
[2622883.492774] SysRq : Show Memory
[2622883.492784] Mem-info:
[2622883.492786] DMA per-cpu:
[2622883.492790] CPU    0: Hot: hi:    0, btch:   1 usd:   0   Cold: hi:    0, btch:   1 usd:   0
[2622883.492793] CPU    1: Hot: hi:    0, btch:   1 usd:   0   Cold: hi:    0, btch:   1 usd:   0
[2622883.492795] DMA32 per-cpu:
[2622883.492797] CPU    0: Hot: hi:  186, btch:  31 usd:  84   Cold: hi:   62, btch:  15 usd:  55
[2622883.492800] CPU    1: Hot: hi:  186, btch:  31 usd: 147   Cold: hi:   62, btch:  15 usd:  56
[2622883.492804] Active:373468 inactive:144663 dirty:0 writeback:28854 unstable:0
[2622883.492806]  free:4600 slab:194244 mapped:12220 pagetables:6339 bounce:0
[2622883.492809] DMA free:8804kB min:16kB low:20kB high:24kB active:0kB inactive:0kB present:8308kB pages_scanned:0 all_unreclaimable? yes
[2622883.492812] lowmem_reserve[]: 0 2988 2988 2988
[2622883.492818] DMA32 free:9596kB min:6984kB low:8728kB high:10476kB active:1493872kB inactive:578652kB present:3060476kB pages_scanned:548 all_unreclaimable? no
[2622883.492821] lowmem_reserve[]: 0 0 0 0
[2622883.492825] DMA: 5*4kB 4*8kB 3*16kB 2*32kB 3*64kB 2*128kB 2*256kB 1*512kB 1*1024kB 1*2048kB 1*4096kB = 8804kB
[2622883.492835] DMA32: 868*4kB 37*8kB 6*16kB 10*32kB 5*64kB 2*128kB 1*256kB 1*512kB 2*1024kB 1*2048kB 0*4096kB = 9624kB
[2622883.492846] Swap cache: add 1758503, delete 1699227, find 18628028/18793383, race 0+120
[2622883.492848] Free swap  = 2530928kB
[2622883.492850] Total swap = 3911784kB
[2622883.492851] Free swap:       2530928kB
[2622883.504163] 780032 pages of RAM
[2622883.504165] 14505 reserved pages
[2622883.504167] 92561 pages shared
[2622883.504168] 59411 pages swap cached

-- 
Do what you love because life is too short for anything else.


[-- Attachment #2: bigmalloc.c --]
[-- Type: text/plain, Size: 476 bytes --]

#include <stdlib.h>
#include <errno.h>
#include <stdio.h>

int main(int argc, char *argv[])
{
	unsigned long sz;
	void *ptr;
	char *endptr;

	if (argc != 2) {
		fprintf(stderr, "usage: %s bytes\n", argv[0]);
		return 1;
	}
	errno = 0;
	sz = strtoul(argv[1], &endptr, 10);
	if (errno || (argv[1] == endptr)) {
		fprintf(stderr, "invalid value\n");
		return 1;
	}
	fprintf(stderr, "trying %lu bytes... ", sz);
	ptr = malloc(sz);
	fprintf(stderr, "got %p\n", ptr);
	return 0;
}

Re: Linux 2.6.24.4+mmap with overcommit >= 1

[Alan Cox]

On Wed, 30 Apr 2008 04:19:54 +0300
Sami Farin <safari-kernel@safari.iki.fi> wrote:

> I have arch x86_64, glibc-2.7.90-7 from Fedora.
> When I have vm.overcommit_memory == 1 or 2,
> mmap succeeds for really huge values.
> These ranges succeed (approximate):
>  8793870000000 - 17500387000000
> 26390387000000 - 35180387000000

Fun fun fun

This should fix strict overcommit (== 2) I think


diff -u --new-file --recursive --exclude-from /usr/src/exclude linux.vanilla-2.6.25-mm1/include/linux/mman.h linux-2.6.25-mm1/include/linux/mman.h
--- linux.vanilla-2.6.25-mm1/include/linux/mman.h	2008-04-28 11:35:21.000000000 +0100
+++ linux-2.6.25-mm1/include/linux/mman.h	2008-04-30 11:21:10.000000000 +0100
@@ -15,16 +15,31 @@
 
 #include <asm/atomic.h>
 
+/* 32bit platforms have a virtual address space in pages we
+   can fit into an atomic_t. We want to avoid atomic64_t on
+   such boxes as it is often expensive and most strict overcommit
+   users turn out to be embedded low power processors */
+
+#if (BITS_PER_LONG == 32)
+#define vm_atomic_t atomic_t
+#define vm_atomic_read atomic_read
+#define vm_atomic_add atomic_add
+#else
+#define vm_atomic_t atomic64_t
+#define vm_atomic_read atomic64_read
+#define vm_atomic_add atomic64_add
+#endif
+
 extern int sysctl_overcommit_memory;
 extern int sysctl_overcommit_ratio;
-extern atomic_t vm_committed_space;
+extern vm_atomic_t vm_committed_space;
 
 #ifdef CONFIG_SMP
 extern void vm_acct_memory(long pages);
 #else
 static inline void vm_acct_memory(long pages)
 {
-	atomic_add(pages, &vm_committed_space);
+	vm_atomic_add(pages, &vm_committed_space);
 }
 #endif
 
diff -u --new-file --recursive --exclude-from /usr/src/exclude linux.vanilla-2.6.25-mm1/mm/mmap.c linux-2.6.25-mm1/mm/mmap.c
--- linux.vanilla-2.6.25-mm1/mm/mmap.c	2008-04-28 11:36:52.000000000 +0100
+++ linux-2.6.25-mm1/mm/mmap.c	2008-04-30 11:17:03.000000000 +0100
@@ -80,7 +80,7 @@
 int sysctl_overcommit_memory = OVERCOMMIT_GUESS;  /* heuristic overcommit */
 int sysctl_overcommit_ratio = 50;	/* default is 50% */
 int sysctl_max_map_count __read_mostly = DEFAULT_MAX_MAP_COUNT;
-atomic_t vm_committed_space = ATOMIC_INIT(0);
+vm_atomic_t vm_committed_space = ATOMIC_INIT(0);
 
 /*
  * Check that a process has enough memory to allocate a new virtual
@@ -177,7 +177,7 @@
 	 * cast `allowed' as a signed long because vm_committed_space
 	 * sometimes has a negative value
 	 */
-	if (atomic_read(&vm_committed_space) < (long)allowed)
+	if (vm_atomic_read(&vm_committed_space) < (long)allowed)
 		return 0;
 error:
 	vm_unacct_memory(pages);
diff -u --new-file --recursive --exclude-from /usr/src/exclude linux.vanilla-2.6.25-mm1/mm/swap.c linux-2.6.25-mm1/mm/swap.c
--- linux.vanilla-2.6.25-mm1/mm/swap.c	2008-04-28 11:36:52.000000000 +0100
+++ linux-2.6.25-mm1/mm/swap.c	2008-04-30 11:18:05.000000000 +0100
@@ -503,7 +503,7 @@
 	local = &__get_cpu_var(committed_space);
 	*local += pages;
 	if (*local > ACCT_THRESHOLD || *local < -ACCT_THRESHOLD) {
-		atomic_add(*local, &vm_committed_space);
+		vm_atomic_add(*local, &vm_committed_space);
 		*local = 0;
 	}
 	preempt_enable();

Re: Linux 2.6.24.4+mmap with overcommit >= 1

[Sami Farin]

On Wed, Apr 30, 2008 at 11:24:53 +0100, Alan Cox wrote:
> On Wed, 30 Apr 2008 04:19:54 +0300
> Sami Farin <safari-kernel@safari.iki.fi> wrote:
> 
> > I have arch x86_64, glibc-2.7.90-7 from Fedora.
> > When I have vm.overcommit_memory == 1 or 2,
> > mmap succeeds for really huge values.
> > These ranges succeed (approximate):
> >  8793870000000 - 17500387000000
> > 26390387000000 - 35180387000000
> 
> Fun fun fun
> 
> This should fix strict overcommit (== 2) I think

Thanks, one more reason to boot to 2.6.24.5...

This patch needed, too, I guess.

diff --git a/fs/proc/proc_misc.c b/fs/proc/proc_misc.c
index 3462bfd..a208a7b 100644
--- a/fs/proc/proc_misc.c
+++ b/fs/proc/proc_misc.c
@@ -132,7 +132,7 @@ static int meminfo_read_proc(char *page, char **start, off_t off,
 #define K(x) ((x) << (PAGE_SHIFT - 10))
 	si_meminfo(&i);
 	si_swapinfo(&i);
-	committed = atomic_read(&vm_committed_space);
+	committed = vm_atomic_read(&vm_committed_space);
 	allowed = ((totalram_pages - hugetlb_total_pages())
 		* sysctl_overcommit_ratio / 100) + total_swap_pages;
 


-- 
Do what you love because life is too short for anything else.

Re: Linux 2.6.24.4+mmap with overcommit >= 1

[Sami Farin]

On Wed, Apr 30, 2008 at 14:47:38 +0300, Sami Farin wrote:
> On Wed, Apr 30, 2008 at 11:24:53 +0100, Alan Cox wrote:
> > On Wed, 30 Apr 2008 04:19:54 +0300
> > Sami Farin <safari-kernel@safari.iki.fi> wrote:
> > 
> > > I have arch x86_64, glibc-2.7.90-7 from Fedora.
> > > When I have vm.overcommit_memory == 1 or 2,
> > > mmap succeeds for really huge values.
> > > These ranges succeed (approximate):
> > >  8793870000000 - 17500387000000
> > > 26390387000000 - 35180387000000
> > 
> > Fun fun fun
> > 
> > This should fix strict overcommit (== 2) I think
> 
> Thanks, one more reason to boot to 2.6.24.5...
> 
> This patch needed, too, I guess.

Booted.  Works great.

$ ./a.out 30390387000000
trying 30390387000000 bytes... got (nil)

-- 
Do what you love because life is too short for anything else.