** Description changed: Binary package hint: linux-image-2.6.24-19-generic Upon switching from Dapper to Hardy, I noticed one of my systems kept getting serial timeouts. Upon further investigation, it seems that there is a serial deadlock issue in kernels < 2.6.27, where machine with multiple processors and multiple serial devices can call the serial interrupt handler at the same time and become deadlocked. This was fixed in 2.6.27 with 4 lines of code. I have created a patch with the same four line based on the linux-source-2.6.24.tbz (from installing linux-source package) and tested it. My problems go away. Given that this is such a fundamental bug (deadlock in interrupt handlers), I feel it is necessary to apply this to the LTS kernels in Hardy. + + The original discussion of this bug and the patch (by the usual kernel programmers) is at: + http://kerneltrap.org/mailarchive/linux-kernel/2008/7/1/2313364 + + Note, that the final fix in 2.6.27 keeps the lock and unlock lines as + they were (the original discussion at the thread above changed them). + My patch does not change them (as in the final 2.6.27 patch for this + problem).
** Description changed: Binary package hint: linux-image-2.6.24-19-generic Upon switching from Dapper to Hardy, I noticed one of my systems kept getting serial timeouts. Upon further investigation, it seems that - there is a serial deadlock issue in kernels < 2.6.27, where machine with - multiple processors and multiple serial devices can call the serial + there is a serial deadlock issue in kernels < 2.6.27, where machines + with multiple processors and multiple serial devices can call the serial interrupt handler at the same time and become deadlocked. This was fixed in 2.6.27 with 4 lines of code. I have created a patch with the same four line based on the linux-source-2.6.24.tbz (from installing - linux-source package) and tested it. My problems go away. Given that - this is such a fundamental bug (deadlock in interrupt handlers), I feel - it is necessary to apply this to the LTS kernels in Hardy. + linux-source package), and I have tested it with many serial deives + talking at once. My problems go away. Given that this is such a + fundamental bug (deadlock in interrupt handlers), I feel it is necessary + to apply this to the LTS kernels in Hardy. The original discussion of this bug and the patch (by the usual kernel programmers) is at: http://kerneltrap.org/mailarchive/linux-kernel/2008/7/1/2313364 Note, that the final fix in 2.6.27 keeps the lock and unlock lines as they were (the original discussion at the thread above changed them). My patch does not change them (as in the final 2.6.27 patch for this problem). -- serial deadlock with SMP https://bugs.launchpad.net/bugs/280821 You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
