memory corruption since 2.1.89: found a possible explanation.

Dario_Ballabio (Dario_Ballabio@milano.europe.dg.com)
Sun, 5 Apr 1998 20:19:37 +0200


I noticed that, starting from 2.1.89, there was an unusual number
of overlapping i/o requests. In average the number of overlapping
requests before 2.1.89 was about 1 every 20000 on my system,
now is in the range of 1 every 20.
Also starting from 2.1.89 I can no longer compile a kernel
due to memory corruption (gcc internal compiler error...)
while the hardware is still pretty stable when using 2.0.33
and I can still compile the 2.1.92 kernel using 2.0.33.
The same hardware (486DX33, 8MB) has been compiling ALL kernels
since 1.1.2, so I can really trust it.

I monitored the overlapping requests and they are directed
to my swap device, here is a typical example:

pid 6089 sector 3472 size 8 write
pid 6090 sector 3488 size 56 write
pid 6091 sector 896 size 8 read
pid 6092 sector 2520 size 8 read
pid 6093 sector 3528 size 8 read

Here pid 6090 and pid 6093 overlap and if for some reason they are
performed in reverse order (think to scsi tagged commands)
this would clearly explain the memory corruption.

+--------------------------------+--------------------------------------+
| Dario Ballabio, | Phone : +39 2 753961 |
| Technical Director, | Fax : +39 2 75396204 |
| Data General Italy | Mob. phone: +39 3356322300 |
| | Addr. : v. Rivoltana, 13 |
| | 20090 Segrate MI Italy |
| | Email: dario@milano.europe.dg.com |
| | |
| | "si vis pacem, para bellum." |
+--------------------------------+--------------------------------------+

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu