Hi.
I did:
echo -20 > /sys/class/firmware/timeout
modprobe ueagle-atm
and the oops was generated.
Module ueagle-atm (http://castet.matthieu.free.fr/eagle/ueagle-pre12.tar.gz),
call request_firmware(). Variable loading_timeout should be probably unsigned,
but this is not direct oops reason. There are races between
firmware_loading_show() and request_firmware() when someone
write 1 to $SYSFS/$DEVPATH/loading file after releasing firmware.
Attached small patch should fix problem.
--
Staszek Gruszka
OOPS:
Unable to handle kernel NULL pointer dereference at virtual address 00000004
printing eip:
d08dd1e9
*pde = 00000000
Oops: 0000 [#1]
PREEMPT SMP
Modules linked in: ueagle_atm usbatm pppoatm ppp_generic slhc firmware_class crc32 nvidia_agp intel_agp clip atm i810_audio ac97_codec soundcore agpgart uhci_hcd usbcore video thermal processor fan button battery ac
CPU: 0
EIP: 0060:[pg0+273338857/1069343744] Not tainted VLI
EFLAGS: 00010246 (2.6.12-rc5)
EIP is at firmware_loading_store+0x89/0xf0 [firmware_class]
eax: 00000000 ebx: 00000001 ecx: 0000000a edx: 00000028
esi: cf0a8680 edi: c3b07fa4 ebp: 080f3408 esp: c3b07f0c
ds: 007b es: 007b ss: 0068
Process firmware.agent (pid: 3577, threadinfo=c3b06000 task=cfc40020)
Stack: c3acf000 00000000 0000000a cabd4100 d08ded04 c0234dfa cabd4100 c3acf000
00000002 c0394404 c01935be cabd4108 d08ded04 c3acf000 00000002 c5e00bc0
c8b13980 c0193628 c78bd8c0 c5e00bc0 00000002 00000002 00000000 c8b13980
Call Trace:
[class_device_attr_store+58/64] class_device_attr_store+0x3a/0x40
[flush_write_buffer+62/80] flush_write_buffer+0x3e/0x50
[sysfs_write_file+88/128] sysfs_write_file+0x58/0x80
[vfs_write+174/304] vfs_write+0xae/0x130
[sys_write+81/128] sys_write+0x51/0x80
[syscall_call+7/11] syscall_call+0x7/0xb
Code: 6e 58 02 8d 46 20 e8 f7 b3 83 ef 8b 44 24 20 8b 5c 24 0c 8b 74 24 10 83 c4 14 c3 f0 ff 0d 24 ec 8d d0 0f 88 18 0a 00 00 8b 46 54 <8b> 40 04 89 04 24 e8 8c 6f 87 ef 8b 46 54 c7 40 04 00 00 00 00
PATCH:
diff -up drivers/base/firmware_class.c{.orig,}
--- drivers/base/firmware_class.c.orig 2005-05-26 22:47:44.000000000 +0200
+++ drivers/base/firmware_class.c 2005-05-26 22:48:21.000000000 +0200
@@ -428,6 +428,7 @@ request_firmware(const struct firmware *
set_bit(FW_STATUS_DONE, &fw_priv->status);
del_timer_sync(&fw_priv->timeout);
+ class_device_unregister(class_dev);
down(&fw_lock);
if (!fw_priv->fw->size || test_bit(FW_STATUS_ABORT, &fw_priv->status)) {
@@ -437,7 +438,6 @@ request_firmware(const struct firmware *
}
fw_priv->fw = NULL;
up(&fw_lock);
- class_device_unregister(class_dev);
goto out;
error_kfree_fw:
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
[Index of Archives]
[Kernel Newbies]
[Netfilter]
[Bugtraq]
[Photo]
[Stuff]
[Gimp]
[Yosemite News]
[MIPS Linux]
[ARM Linux]
[Linux Security]
[Linux RAID]
[Video 4 Linux]
[Linux for the blind]
[Linux Resources]