[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnash-commit] [SCM] Gnash branch, master, updated. release_0_8_9_fi
From: |
John Gilmore |
Subject: |
Re: [Gnash-commit] [SCM] Gnash branch, master, updated. release_0_8_9_final-1963-g63de409 |
Date: |
Sun, 27 Apr 2014 19:44:38 -0700 |
Hi Bastiaan!
> Savannah #42199: avoid thrashing the stack in case of a high fd by
> using poll() instead of select().
It looks like the variable <fd> has been trashed, since the second
backtrace shows a nine-digit file descriptor that the kernel clearly
would never implement. It would be simpler to add a check for "if (fd
> FD_SETSIZE) abort();". -- though it looks like we crash in this case
anyway, so we don't really need the check.
What we need is to figure out why a bogus <fd> was passed in to this
routine.
I recommend reverting the change, since it didn't actually fix a bug nor
contribute to fixing it. Also, the change introduced an extraneous change:
the original timeout was 2 seconds and the poll changes it to 1.
John