[Kernel-packages] [Bug 1500954] [NEW] xfs DIO data corruption

2015-09-29 Thread Avi Kivity
Public bug reported: An appending DIO write will corrupt files on XFS. The following patch needs to be backported: commit b9d59846f73713d77f0f3fb784c7f84249fc2b93 Author: Dave Chinner Date: Thu Apr 16 22:03:07 2015 +1000 xfs: DIO write completion size updates race xfs_end_io_dir

[Kernel-packages] [Bug 1500954] Re: xfs DIO data corruption

2015-10-15 Thread Avi Kivity
Any progress on this issue? This is DATA CORRUPTION! -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux-lts-vivid in Ubuntu. https://bugs.launchpad.net/bugs/1500954 Title: xfs DIO data corruption Status in linux-lts-vivid package i

[Kernel-packages] [Bug 1500954] Re: xfs DIO data corruption

2015-10-17 Thread Avi Kivity
** Package changed: linux-lts-vivid (Ubuntu) => linux-lts-trusty (Ubuntu) -- You received this bug notification because you are a member of Kernel Packages, which is subscribed to linux-lts-trusty in Ubuntu. https://bugs.launchpad.net/bugs/1500954 Title: xfs DIO data corruption Status in linu

[Kernel-packages] [Bug 2084384] [NEW] CONFIG_READ_ONLY_THP_FOR_FS is not set

2024-10-13 Thread Avi Kivity
Public bug reported: CONFIG_READ_ONLY_THP_FOR_FS allows transparent hugepages to apply to .text segments of executables (if they execute the appropriate madvise() call), leading to better performance. Unfortunately, it is not enabled in Ubuntu. Please enable it. Verified in 24.10: # grep CONFIG

[Kernel-packages] [Bug 2084384] Re: CONFIG_READ_ONLY_THP_FOR_FS is not set

2024-10-28 Thread Avi Kivity
Yes, that's the option. Note that the application needs to opt in via madvise(). Here's a case study from Intel: https://www.intel.com/content/dam/develop/external/us/en/documents/runtimeperformanceoptimizationblueprint- largecodepages-q1update.pdf -- You received this bug notification because y