Messages in this thread | | | Date | Sat, 19 Jul 1997 23:46:05 +0200 | From | Philippe Strauss <> | Subject | Buffer-Werner-17-jul: Deadlock again.. |
| |
Yeap, my machine just frozen while compiling pre-2.1.46-1. I'm using 2.0.30 + pre31-2, with Werner's buffer patch of july 17 _without_ removing
> /* Give the physical reallocated page a bigger start */ > mem_map[MAP_NR(page)].age = (2*PAGE_INITIAL_AGE);
this. Total uptime with this kernel (or any kernel with this buffer patch) was 5 hours approx.
Complete list of patch in this kernel:
1 ./ 0 ext2-error 0 pre31-2 1 ../ 0 filehandle 0 stack-exec 0 buff-werner-17-jul 0 ft
I also got this message in the logs from this kernel, way before the crash, while dumping a large file from a scsi dat tape with 'dd if=/dev/nst0 of=file bs=32k'
Jul 19 15:10:19 lili kernel: try_to_free_page: free pages 932kB, async pages 0kB
When the crash occured, i was on a the VC doing the compile, with X running and eating its normal (huge) memory requirement.
My hardware setup is P5 166, 430FX, 64MB RAM, ncr53c810 scsi adapter and ncrBsd driver. NOTE: my RAM setup is the following: 2 matched simms bought one year ago (2x16M) and 2 more matched simms bought last week. I know (by personal experience with low quality and unmatched pair of simms) that this kind of setup _may_ lead to problem. Anyway, I run this setup since 7 days, with a lot of compilation, without a problem. But without ECC, I can't be sure that it's not hardware related. Seeing Miquel's news server crashing in the same (silent) way, I think this is more likely software related.
Software setup is development debian of yesterday (debian hamm)
Cheers.
-- Philippe Strauss <philou@lili.urbanet.ch>
Homepage & PGP key: http://lili.urbanet.ch
"I know not with what weapons World War III will be fought, but World War IV will be fought with sticks and stones." -- Albert Einstein --
| |