fds jumping between processes
Hello,
here is quite an open bug report... I could not reproduce this again, but it involves the fds opened of a killed process being passed to another process, or something like this.
I'm using vanilla 2.6.33.3 in x86_64.
I was using skype, talking through it using alsa, and trying the 'share desktop' feature, it died. Starting it again, it could not open the alsa device. Using fuser:
...:/dev/snd]$ fuser * controlC0: 32341 pcmC0D0p: 32341 timer: 32341
Which that was:
$ ps ax | grep 323 32341 ? S 0:01 /home/xxx/.nix-profile/bin/xpdf /tmp/xxx.pdf
So, after skype (using alsa) died, the xpdf process took the soundcard fds skype had. It does not make much sense xpdf to use sound... so I guess something worth looking happened.
I don't have more details. Killing xpdf made the new skype open the alsa device properly.
Sorry for not having analised the situation better. Now I cannot reproduce it anymore. Maybe this will make someone knowing say "ah, let me look if ...". I just thought if this is a bug in the kernel, it can involve security problems.
Thank you,
Lluís.
Bah, forget it! firefox problem. Sorry for the noise.
Quoted 32 lines · Lluís Batlle i Rossell
Hello,
here is quite an open bug report... I could not reproduce this again, but it involves the fds opened of a killed process being passed to another process, or something like this.I'm using vanilla 2.6.33.3 in
x86_64.
I was using skype, talking through it using alsa, and trying the 'share desktop' feature, it died. Starting it again, it could not open the alsa device. Using fuser:...:/dev/snd]$ fuser * controlC0: 32341 pcmC0D0p: 32341 timer: 32341Which that was:
$ ps ax | grep 323 32341 ? S 0:01 /home/xxx/.nix-profile/bin/xpdf /tmp/xxx.pdfSo, after skype (using alsa) died, the xpdf process took the soundcard fds skype had. It does not make much sense xpdf to use sound... so I guess something worth looking happened.
I don't have more details. Killing xpdf made the new skype open the alsa device properly.
Sorry for not having analised the situation better. Now I cannot reproduce it anymore. Maybe this will make someone knowing say "ah, let me look if ...". I just thought if this is a bug in the kernel, it can involve security problems.
Thank you,
Lluís.