[Click] [click] occuring segmentation fault in cilck-install

[Eddie Kohler]

Hi Jason,

I think this is a report of a potential bug, not an actual bug.  I've checked 
in a fix: use GFP_ATOMIC in the kmalloc()s in proclikefs, since they might be 
called when a lock is held.

Eddie


Jason Haas wrote:
> I actually have a vanilla 2.6.13.2, patched kernel running a CVS version
> downloaded on Jan. 17th.  It's a Fedora Core 3 machine.  I have two other
> machines (both vanilla 2.6.13.2 + click patch), one SuSE 9.1-pro and one FC4
> that don't work.  The funny thing is, that even though there are errors on
> insertion, the configurations run correctly, but upon trying to uninstall
> them, the kernel seg faults.  I've tried copying the .config for the kernel
> from the working machine to one of the broken ones, but it doesn't boot
> then.
> 
> Here's the relevant dmesg for insert (from running test.click):
> 
> mtrr: base(0xe8020000) is not aligned on a size(0x800000) boundary
> click: no version for "proclikefs_new_inode_operations" found: kernel
> tainted.
> Debug: sleeping function called from invalid context at mm/slab.c:2096
> in_atomic():1, irqs_disabled():0
>  [<c0104256>] dump_stack+0x17/0x19
>  [<c0123166>] __might_sleep+0x9c/0xaa
>  [<c01652a6>] __kmalloc+0xcf/0xd9
>  [<f8962f62>] proclikefs_register_filesystem+0x3a0/0x52f [proclikefs]
>  [<fb118583>] _Z12init_clickfsv+0x17/0x1aa [click]
>  [<fb118392>] init_module+0x354/0x382 [click]
>  [<c0150233>] sys_init_module+0xcb/0x24f
>  [<c0103add>] syscall_call+0x7/0xb
> click: starting router thread pid 8625 (f78ddc5c)
> chatter: ok:   40 | 45000028 00000000 401177c3 01000001 02000002 13691369
> chatter: ok:   40 | 45000028 00000000 401177c3 01000001 02000002 13691369
> chatter: ok:   40 | 45000028 00000000 401177c3 01000001 02000002 13691369
> chatter: ok:   40 | 45000028 00000000 401177c3 01000001 02000002 13691369
> chatter: ok:   40 | 45000028 00000000 401177c3 01000001 02000002 13691369
> 
> Here's the uninstall report:
> 
> Debug: sleeping function called from invalid context at
> /usr/src/linux/include/asm/semaphore.h:102
> in_atomic():1, irqs_disabled():0
>  [<c0104256>] dump_stack+0x17/0x19
>  [<c0123166>] __might_sleep+0x9c/0xaa
>  [<f89634e2>] proclikefs_unregister_filesystem+0x3f1/0xbe4 [proclikefs]
>  [<f8cb7d7f>] _Z15cleanup_clickfsv+0x29/0x232 [click]
>  [<f8cb7071>] cleanup_module+0xb/0xbc [click]
>  [<c014def4>] sys_delete_module+0x117/0x155
>  [<c0103add>] syscall_call+0x7/0xb
> click: stopping router thread pid 27961
> chatter: deleting router while ref count = 1802201963
> Unable to handle kernel paging request at virtual address 6b6b6b73
>  printing eip:
> f8c04564
> *pde = 00000000
> Oops: 0002 [#1]
> PREEMPT
> Modules linked in: click proclikefs rt2570 parport_pc lp parport autofs4
> rfcomm l2cap bluetooth video button battery ac ipv6 i2c_i801 i2c_core
> snd_intel8x0 snd_ac97_codec snd_seq_dummy snd_seq_oss snd_seq_midi_event
> snd_seq snd_seq_device snd_pcm_oss snd_mixer_oss snd_pcm snd_timer snd
> snd_page_alloc e100 floppy
> CPU:    0
> EIP:    0060:[<f8c04564>]    Tainted: GF     VLI
> EFLAGS: 00210202   (2.6.13.2-click)
> EIP is at _ZN6Router5unuseEv+0x6/0x26 [click]
> eax: 6b6b6b6b   ebx: 6b6b6b6b   ecx: ebede0ac   edx: 0000009d
> esi: 00000003   edi: bfff3934   ebp: f66a5f04   esp: f66a5f00
> ds: 007b   es: 007b   ss: 0068
> Process rmmod (pid: 27995, threadinfo=f66a4000 task=eaf4baa0)
> Stack: ebede0ac f66a5f34 f8c04129 f8ce9bb0 6b6b6b6b 00200286 f8d76000
> 00000003
>        ebede0ac f66a5f2c ebede0ac 00000003 bfff3934 f66a5f58 f8cb632b
> eb6d8d84
>        ee773990 bfff3934 f66a5f58 f8cb54f2 f8d76000 00000003 f66a5f68
> f8cb708a
> Call Trace:
>  [<c0104209>] show_stack+0x98/0xce
>  [<c01043e3>] show_registers+0x18b/0x225
>  [<c010464d>] die+0x146/0x2cf
>  [<c011bcb6>] do_page_fault+0x235/0x68d
>  [<c0103cf7>] error_code+0x4f/0x54
>  [<f8c04129>] _ZN6RouterD1Ev+0x2b/0x460 [click]
>  [<f8cb632b>] _Z19click_cleanup_schedv+0x45/0x2a0 [click]
>  [<f8cb708a>] cleanup_module+0x24/0xbc [click]
>  [<c014def4>] sys_delete_module+0x117/0x155
>  [<c0103add>] syscall_call+0x7/0xb
> Code: ff ff 8b 83 88 01 00 00 e8 28 c2 51 c7 e9 e8 fc ff ff e8 9e 16 85 c7
> 8b 4d ec 8b 91 40 01 00 00 e9 fa fc ff ff 55 89 e5 53 89 c3 <ff> 48 08
> 0f 94 c0 84 c0 74 12 85 db 74 0e 89 d8 e8 85 fb ff ff
> 
> I too would appreciate any help, if anyone knows why one might work and
> another not.
> 
> Thanks,
> 
> Jason
> _______________________________________________
> click mailing list
> click at amsterdam.lcs.mit.edu
> https://amsterdam.lcs.mit.edu/mailman/listinfo/click