Skip to contentSkip to conversation

fix oops when reading /proc/ioports

PATCH New Igor Popik 1 participant 1 message Oct 5, 2005 Pcmcia

lore
Share & tools
lore

Igor Popik

author
#0 lore raw

PATCH 1/1 fix oops when reading /proc/ioports

Hi,

I have noticed that in recent kernels (2.6.13 and newer) I got an oops every time I tried to read /proc/ioports after trying to load i82365 pcmcia driver module:

root@alien2:~# modprobe i82365
FATAL: Error inserting i82365
(/lib/modules/2.6.14-rc3/kernel/drivers/pcmcia/i82365.ko): No such
device
root@alien2:~# cat /proc/ioports
<1>Unable to handle kernel paging request at virtual address cff10802
printing eip:
c01c09bc
*pde = 0eec3067
*pte = 00000000
Oops: 0000 [#4]
Modules linked in: rsrc_nonstatic pcmcia_core snd_via82xx gameport
snd_ac97_codec snd_ac97_bus snd_pcm snd_timer
snd_page_alloc
snd_mpu401_uart snd_rawmidi snd soundcore psmouse
CPU:    0
EIP:    0060:[<c01c09bc>]    Not tainted VLI
EFLAGS: 00010297   (2.6.14-rc3)
EIP is at vsnprintf+0x33c/0x4d0
eax: cff10802   ebx: 0000000a   ecx: cff10802   edx: fffffffe
esi: ce525122   edi: 00000000   ebp: ce525fff   esp: c70e5eb4
ds: 007b   es: 007b   ss: 0068
Process cat (pid: 3365, threadinfo=c70e4000 task=c13b1030)
Stack: ce52511b ce525fff 000003e1 00000000 00000010 00000004 00000002 00000001
    ffffffff ffffffff c30747e0 c7f6efc0 c30747e0 00000116 c01723d5 
ce525116
    00000eea c0333dc8 c70e5f2c 00000004 c011cc87 c30747e0 c0333db6 
00000000
Call Trace:
[<c01723d5>] seq_printf+0x35/0x60
[<c011cc87>] r_show+0x77/0x80
[<c0171ed6>] seq_read+0x1d6/0x2d0
[<c01521a7>] vfs_read+0xa7/0x180
[<c0152561>] sys_read+0x51/0x80
[<c0103095>] syscall_call+0x7/0xb
Code: 24 08 00 00 00 83 cf 01 eb bb 8b 44 24 48 8b 54 24 20 83 44 24
48 04 8b 08 b8 ac ed 33 c0 81 f9 ff 0f 00 00 0f 46 c8
89 c8 eb 06 <80> 38 00 74 07 40 4a 83 fa ff 75 f4 29 c8 83 e7 10 89 c3 75 20
Segmentation fault

This oops is caused by broken i82365 PCMCIA driver. Someone tried to remove deprecated check_region() but after the changes the driver does not release its memory regions correctly.

I've attached patch that fixes that oops. This driver looks really dirty, and I do not know if it still works.

Cheers,
Igor

drivers/pcmcia/i82365.c 2005-10-05 00:07:04.000000000 +0200 renamed from drivers/pcmcia/i82365.c 2005-08-29 01:41:01.000000000 +0200 +7 -2
@@ -697,6 +697,11 @@
  struct i82365_socket *t = &socket[sockets-ns];
 
  base = sockets-ns;
+    if (t->ioaddr > 0) {
+       if (!request_region(t->ioaddr, 2, "i82365"))
+           printk(KERN_WARNING "port conflict at %#lx\n", t->ioaddr);
+    }
+
  if (base == 0) printk("\n");
  printk(KERN_INFO "  %s", pcic[type].name);
  printk(" ISA-to-PCMCIA at port %#lx ofs 0x%02x",
@@ -806,7 +811,8 @@
     if (sockets == 0)
         printk("port conflict at %#lx\n", i365_base);
     return;
-    }
+    } else
+       release_region(i365_base, 2);
 
  id = identify(i365_base, 0);
  if ((id == IS_I82365DF) && (identify(i365_base, 1) != id)) {
@@ -1440,7 +1446,6 @@
     i365_set(i, I365_CSCINT, 0);
     release_region(socket[i].ioaddr, 2);
  }
-    release_region(i365_base, 2);

#ifdef CONFIG_PNP
if (i82365_pnpdev)

             pnp_disable_dev(i82365_pnpdev);