Changes between Initial Version and Version 1 of Ticket #256, comment 3


Ignore:
Timestamp:
Mar 12, 2012, 3:19:53 PM (12 years ago)
Author:
dmik
Comment:

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #256, comment 3

    initial v1  
    1 I actually checked glibc already. Although they define a cleanup routine that removes the lock but it's not intended for signal handlers it seems. However I don't think you can fully use Linux as a reference here since it doesn't have the thread 1 exitlist hangup problem at all...
     1I actually checked glibc already. Although they define a cleanup routine that removes the lock, it's not intended for signal handlers it seems. However I don't think you can fully use Linux as a reference here since it doesn't have the thread 1 exitlist hangup problem at all...
    22
    33Regarding the bad state. I don't see how my patch changes the situation here. It only comes into play when the thread gets killed, but when it happens the state of everything (including streams) is potentially messed up anyway, regardless of what I changed. My change is only to prevent a loud annoying crash or a much more annoying hang leading to an unkillable zombie.