lkml.org 
[lkml]   [2000]   [Feb]   [1]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectIDE weird behavior in 2.3.41 w/ synchronous FTP reception

Hello,

I'm facing a problem in FTP reception problem with a new synchronous card
driver (the Cyclades-PC300 driver), but I've just found out something at
least interesting.

If I do FTP _transmissions_ , no problems happen.

If I do FTP _receptions_ , at high speeds (e.g. 4Mbps) and using big files
(e.g. 10MB) the system crashes hard, with an oops report that's evidently
corrupted:

Unable to handle kernel paging request at virtual address 0d841801
printing eip:
0d841801
*pde = 00000000
Oops: 0000
CPU: 0
EIP: 0010:[<0d841801>]
EFLAGS: 00010286
eax: 00000009 ebx: c289c048 ecx: 00000001 edx: 00000000
esi: c28173e0 edi: c2816bbc ebp: 00000000 esp: c027bed0
ds: 0018 es: 0018 ss: 0018
Process swapper (pid: 0, stackpage=c027b000)
Stack: d4ae2b59 c90e0f7d 1bde679c 27c3302c eaf3f1b9 9fcc90be 21aa8c63
238d6b20
87823b9e 5742b078 8f534e42 2252c054 67c34420 a0ad1633 9e6388da
f285c016
04c120da e1f2d7c2 88cbed94 14fdf40f e9d9dacd 2fee179e 4f91e7b0
a4174c2a
Call Trace:
Code: <1>Unable to handle kernel paging request at virtual address
0d841801
printing eip:
c010bb3a
*pde = 00000000

The stack has really weird values, and no call / code trace is reported.


Now, the interesting part: if I do FTP receptions _but_ I redirect the
incoming file to /dev/null (i.e., I don't write the file to the IDE HD,
just receive it through the line and save it as /dev/null), no crashes
happen whatsoever !!!

Also, if I replace the IDE HD by a SCSI HD, running in the same system, no
crashes happen whatsoever.

I've tested this in both a P166 / 32MB RAM and in a Dual PII-400 / 128MB
RAM. In the Dual PII, the problem takes longer to happen, but nevertheless
it does happen.

Furthermore, I tested it in kernels 2.2.14 / 2.2.15pre5, and the problem
does NOT happen. It happens in kernels 2.3.40 / 2.3.40pre4 / 2.3.41.

Again:
- The problem does not happen in the latest 2.2.x kernels.
- The problem _always_ happens when the IDE HD is used and the file from
the FTP reception transfer is written to the HD.
- The problem _never_ happens when the IDE HD is used and the file from
the FTP reception transfer is written to /dev/null.
- The problem _never_ happens when the SCSI HD is used in any test
scenario (i.e., writing or not to the HD).

Can anybody give some suggestions on where I should look at in order to
try and solve this problem?? Any help would be highly apreciated. If you
need to look at the PC300 driver source code, please let me know and I can
send it to you.

TIA and Regards,
Ivan


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:56    [from the cache]
©2003-2011 Jasper Spaans