Re: dma(8): Fix race condition in multi-recipient delivery
:Matthew Dillon wrote:
:> I believe flock() behaves more rationally.
:
:Yah, I simply went and re-opened the file instead of using locks. I also
:have a new queueing code in the works that will get rid of this problem
:entirely.
:
:cheers
: simon
The last time I looked at the code I also noticed that the file
descriptors were getting spammed by the fork()... that the children
would inherit the parent's entire fd set, even open fd's related
to queue files other then the one the child is supposed to
process.
-Matt
討論串 (同標題文章)
完整討論串 (本文為第 11 之 11 篇)：