fails with SIGSEGV

[prspkt]

After a quick build on alpine linux, slurm seems to consistently segfault during runtime.
Stepping through with gdb the error happens here with BUFSIZ - 1 being greater than BUFSIZE.

Applying this patch seems to fix the issue.

[mattthias]

Hi,

the answer might come a year to late :-)

BUFSIZ is a macro that is defined in stdio.h (glibc 8192 / musl 1024) and BUFSIZE is set to 256 in src/linux.h.

The fgets tries to read max BUFSIZ-1 (on my amd64 with glibc 8191) chars into a buffer that is only BUFSIZE (256) chars big.

That indeed looks wrong. I was wondering how this could ever work but looking at the first two lines of /proc/net/dev i see that this lines only ~125 chars long. If your slurm really crashes on this code line i would like to see the content of /proc/net/dev.
But in the actual interface / data lines i can imagine that we can reach a line lengths of more then 256 chars.

[klockeph]

I just installed slurm on arch linux from the AUR (https://aur.archlinux.org/packages/slurm/) and I experience a similar behaviour; it segfaults after a few seconds of runtime.
While I did not investigate with gdb yet, I imagine it could be the same issue. my proc/net/dev looks like the following

Inter-|   Receive                                                |  Transmit
 face |bytes    packets errs drop fifo frame compressed multicast|bytes    packets errs drop fifo colls carrier compressed
    lo: 391504450597 8543650    0    0    0     0          0         0 391504450597 8543650    0    0    0     0       0          0
enp2s0f0:       0       0    0    0    0     0          0         0        0       0    0    0    0     0       0          0
wlp3s0: 132889587  223204    0  529    0     0          0         0 32549645   67318    0    0    0     0       0          0
docker0: 144529703   98316    0    0    0     0          0         0 337031942   58305    0    0    0     0       0          0
veth6a04989: 145906127   98316    0    0    0     0          0         0 337033088   58320    0    0    0     0       0          0

[mattthias]

Hello @klockeph & @prspkt ,

the issue was solved by this 43ed060

Thanks for your report (even i have a horrible lag in answering / fixing stuff).