Summary: | sysstat 5.0.5-r1's iostat strange behaviour | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | barthek <gejzer> |
Component: | New packages | Assignee: | Tim Yamin (RETIRED) <plasmaroo> |
Status: | RESOLVED UPSTREAM | ||
Severity: | normal | ||
Priority: | High | ||
Version: | 2005.0 | ||
Hardware: | All | ||
OS: | All | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
barthek
2005-05-10 12:52:40 UTC
Looks like an upstream bug - please file one with the author upstream. Also you might want to try an older version of sysstat and try without -O3 and -s. fortunately it started to work find with -O2 instead of -O3. weird i havent thought about it before ? the funny thing is that it worked with O3 on 2.4 kernel and on 2.6 it wouldnt. could you please elaborate a little about why it would work while ptraced (strace/gdb) and wouldnt without it ? argh i was too quick about reporting. take a look at this and try to exlain :) doli@1g ~ $ strace -o /dev/null iostat -x 1 1 Linux 2.6.11-gentoo-r8 (1g) 05/16/05 avg-cpu: %user %nice %sys %iowait %idle 16.18 72.58 10.03 0.56 0.64 Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util hda 0.60 3.53 2.95 3.09 39.49 52.98 19.75 26.49 15.31 1.14 188.27 8.91 5.38 hdc 0.24 18.18 6.80 5.73 268.38 191.32 134.19 95.66 36.69 0.50 39.52 5.05 6.32 doli@1g ~ $ iostat -x 1 1 Linux 2.6.11-gentoo-r8 (1g) 05/16/05 avg-cpu: %user %nice %sys %iowait %idle 16.18 72.58 10.03 0.56 0.64 Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util doli@1g ~ $ su - Password: 1g ~ # iostat -x 1 1 Linux 2.6.11-gentoo-r8 (1g) 05/16/05 avg-cpu: %user %nice %sys %iowait %idle 16.18 72.58 10.03 0.56 0.64 Device: rrqm/s wrqm/s r/s w/s rsec/s wsec/s rkB/s wkB/s avgrq-sz avgqu-sz await svctm %util hda 0.60 3.53 2.95 3.09 39.50 52.98 19.75 26.49 15.31 1.14 188.27 8.91 5.38 hdc 0.24 18.18 6.80 5.73 268.43 191.32 134.21 95.66 36.69 0.50 39.52 5.05 6.32 1g ~ # any thoughts on this? Nope, this looks like an upstream bug. However, I recommend you try compiling it with no CFLAGS (including no stack smashing protection) and see if that solves the issue. If it does, reopen this bug and then we know it's a possible toolchain issue. |