From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.1) Gecko/20020827 Description of problem: I recompiled the 2.4.20-2.2{2,6,7} kernels on a up2date RHL 7.3 from the kernel-source rpm using the athlon config from the configs directory. Starting wine on the 2.4.20-2.2{6,7} kernel dosn't work; it just hangs without spitting out the normal warnings it spits out on startup. The wine process can't be terminated by CTRL-C nor moved into background. killall and top just hang and ps outputs the details for the processes with pid smaller then the pid of the wine process and then hangs too. Before starting wine ps/top/killall just work fine. The 2.4.20-2.22 kernel works just fine without exhibiting any of the above problems. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. boot a 2.4.20-2.2{6,7} kernel 2. start wine 3. start top/ps/killall Actual Results: wine, top, ps ,killall hangs Expected Results: they should not hang Additional info: wine is latest CVS but that shouldn't matter. I'll attach the output of Sysrq-t
Created attachment 89663 [details] output of Sysreq-t after the wine/ps/top/killall process locked up
Note: may be related to FASTCALL macros interacting with gcc 2.96. Can remove them for older gcc.
Created attachment 89668 [details] remove FASTCALL from vm_ in lnclude/linux/mm.h As requested i tried this patch without succes.
Created attachment 89669 [details] sysreq-t output for the 2.4.20-2.27 with the patch applied
Forgot to say that i did a "make clean" after applying the patch
I am seeing the killall problem on phoebe2+rawhide packages for kernel -2.27 when the shutdown stops at the "sending TERM signal to all processes" stage. The -2.25 kernel works fine.
it looks like -2.30 works ok here... should be in rawhide now.
Just tested it and wine works just fine. You can close the bug.
So the signal fixes in 2.30 did the trick - great.