Skip to contentSkip to conversation

/proc/*/pagemap oopses

PATCH In review Eric Sesterhenn 3 participants 5 messages Jul 16, 2008

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

hi,

with current -git a "cat /proc/4260/pagemap"
gives me the following oops:

[   80.334016] general protection fault: fffc [#1] PREEMPT DEBUG_PAGEALLOC
[   80.334402] Modules linked in:
[   80.334654] 
[   80.334765] Pid: 4744, comm: cat Not tainted (2.6.26-02962-g4515889 #5)
[   80.334902] EIP: 0060:[<cb613f4b>] EFLAGS: 00010206 CPU: 0
[   80.335093] EIP is at 0xcb613f4b
[   80.335213] EAX: 96c26000 EBX: c013b06a ECX: 00400000 EDX: 00000000
[   80.335347] ESI: cb613f44 EDI: fffffff2 EBP: c016d99c ESP: cb613f04
[   80.335481]  DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
[   80.335672] Process cat (pid: 4744, ti=cb613000 task=cb59df40 task.ti=cb613000)
[   80.335810] Stack: ffffffb0 bfffffff c0000000 00000000 cbb67000 00400000 00000001 00000000 
[   80.336019]        fffffff2 cb613f74 c01abab0 0828a000 0828a400 cb408380 cb400fe0 cb4a6c80 
[   80.336019]        c013b06a cb613f60 c01ab441 00000000 c01ab3af cb4a6c80 cb613f60 0828a000 
[   80.336019] Call Trace:
[   80.336019]  [<c01abab0>] ? pagemap_read+0x16a/0x1e3
[   80.336019]  [<c013b06a>] ? put_lock_stats+0xd/0x21
[   80.336019]  [<c01ab441>] ? pagemap_pte_range+0x0/0xdc
[   80.336019]  [<c01ab3af>] ? pagemap_pte_hole+0x0/0x37
[   80.336019]  [<c01ab946>] ? pagemap_read+0x0/0x1e3
[   80.336019]  [<c017af1f>] ? vfs_read+0x8a/0x106
[   80.336019]  [<c017b21f>] ? sys_read+0x3b/0x60
[   80.336019]  [<c01038b1>] ? sysenter_past_esp+0x6a/0xa5
[   80.336019]  =======================
[   80.336019] Code: 00 00 00 00 f2 ff ff ff 74 3f 61 cb b0 ba 1a c0 00
a0 28 08 00 a4 28 08 80 83 40 cb e0 0f 40 cb 80 6c 4a cb 6a b0 13 c0 60
3f 61 <cb> 41 b4 1a c0 00 00 00 00 af b3 1a c0 80 6c 4a cb 60 3f 61 cb 
[   80.336019] EIP: [<cb613f4b>] 0xcb613f4b SS:ESP 0068:cb613f04
[   80.455182] ---[ end trace e2b2d610b672cbbc ]---

doing this again gives a slightly different oops

[  156.814961] BUG: unable to handle kernel NULL pointer dereference at 00000002
[  156.815288] IP: [<00000002>]
[  156.815482] *pde = 00000000 
[  156.815704] Oops: 0000 [#2] PREEMPT DEBUG_PAGEALLOC
[  156.816079] Modules linked in:
[  156.816255] 
[  156.816368] Pid: 4765, comm: cat Tainted: G      D (2.6.26-02962-g4515889 #5)
[  156.816608] EIP: 0060:[<00000002>] EFLAGS: 00010202 CPU: 0
[  156.816739] EIP is at 0x2
[  156.816854] EAX: cbb67000 EBX: 00000002 ECX: 00400000 EDX: 00000000
[  156.816989] ESI: cb61ef44 EDI: fffffff2 EBP: cb61ef28 ESP: cb61ef00
[  156.817183]  DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
[  156.817314] Process cat (pid: 4765, ti=cb61e000 task=cb626f20 task.ti=cb61e000)
[  156.817455] Stack: c016d99c cb61ef44 bfffffff c0000000 00000000 cbb67000 00400000 00000001 
[  156.818028]        00000000 fffffff2 cb61ef74 c01abab0 09740000 09740400 cb4083b8 cb400fe0 
[  156.818028]        cb4a6c80 00000002 00000000 c01ab441 00000001 c01ab3af cb4a6c80 cb61ef60 
[  156.818028] Call Trace:
[  156.818028]  [<c016d99c>] ? walk_page_range+0x70/0x180
[  156.818028]  [<c01abab0>] ? pagemap_read+0x16a/0x1e3
[  156.818028]  [<c01ab441>] ? pagemap_pte_range+0x0/0xdc
[  156.818028]  [<c01ab3af>] ? pagemap_pte_hole+0x0/0x37
[  156.818028]  [<c01ab946>] ? pagemap_read+0x0/0x1e3
[  156.818028]  [<c017af1f>] ? vfs_read+0x8a/0x106
[  156.818028]  [<c017b21f>] ? sys_read+0x3b/0x60
[  156.818028]  [<c01038b1>] ? sysenter_past_esp+0x6a/0xa5
[  156.818028]  =======================
[  156.818028] Code:  Bad EIP value.
[  156.818028] EIP: [<00000002>] 0x2 SS:ESP 0068:cb61ef00
[  156.823465] ---[ end trace e2b2d610b672cbbc ]---

Greetings, Eric

#1 lore raw
Quoted 2 lines · Eric Sesterhenn

with current -git a "cat /proc/4260/pagemap"
gives me the following oops:

reproduced

#2 lore raw

PATCH 1/1 proc: fix /proc/*/pagemap some more

Quoted 5 lines · Alexey Dobriyan
Quoted 2 lines · Eric Sesterhenn

with current -git a "cat /proc/4260/pagemap"
gives me the following oops:

reproduced

Oh, wow!

[PATCH] proc: fix /proc/*/pagemap some more

struct pagemap_walk was placed on stack, some hooks are initialized, the rest (->pgd_entry, ->pud_entry, ->pte_entry) are valid but junk.

Reported by Eric Sesterhenn.

---

 fs/proc/task_mmu.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/proc/task_mmu.c +1 -1
@@ -636,7 +636,7 @@ static ssize_t pagemap_read(struct file *file, char __user *buf,
 	struct pagemapread pm;
 	int pagecount;
 	int ret = -ESRCH;
-	struct mm_walk pagemap_walk;
+	struct mm_walk pagemap_walk = {};
 	unsigned long src;
 	unsigned long svpfn;
 	unsigned long start_vaddr;
#3 lore raw

Re: [PATCH] proc: fix /proc/*/pagemap some more

Quoted 10 lines · Alexey Dobriyan
Quoted 5 lines · Alexey Dobriyan
Quoted 2 lines · Eric Sesterhenn

with current -git a "cat /proc/4260/pagemap"
gives me the following oops:

reproduced

Oh, wow!

[PATCH] proc: fix /proc/*/pagemap some more

This reproduced for me on plain v2.6.26 as well. I'm guessing this should be Cc stable as soon as possible...!

Vegard

Signature · 4 lines
"The animistic metaphor of the bug that maliciously sneaked in while
the programmer was not looking is intellectually dishonest as it
disguises that the error is the programmer's own creation."
	-- E. W. Dijkstra, EWD1036
#4 lore raw

Re: [PATCH] proc: fix /proc/*/pagemap some more

In reply to Alexey Dobriyan
Quoted 10 lines · * Alexey Dobriyan ([email protected]) wrote:
Quoted 5 lines · Alexey Dobriyan
Quoted 2 lines · Eric Sesterhenn

with current -git a "cat /proc/4260/pagemap"
gives me the following oops:

reproduced

Oh, wow!

[PATCH] proc: fix /proc/*/pagemap some more

thanks, the patch fixes the issue for me

Greetings, Eric