lkml

LKML News v5.9-rc8

[RFC PATCH v2 00/30] 1GB PUD THP support on x86_64 (Zi Yan) https://lkml.kernel.org/r/20200928175428.4110504-1-zi.yan@sent.com This patch adds support of 1GB THP using PUD. [PATCH v4 00/11] KFENCE: A low-overhead sampling-based memory safety error detector (Marco Elver) https://lkml.kernel.org/r/20200929133814.2834621-1-elver@google.com The 4th version of the KFENCE patchset. [PATCH 0/5] Speed up mremap on large regions (Kalesh Singh) https://lkml.kernel.org/r/20200930222130.4175584-1-kaleshsingh@google.com This patchset optimizes mremap() by allowing memory regions to be able to mapped at PMD/PUD. [PATCH] selftests/vm: 10x speedup for hmm-tests https://lkml.

LKML News v5.9-rc7

[PATCH v3 00/10] KFENCE: A low-overhead sampling-based memory safety error detector (Marco Elver) https://lkml.kernel.org/r/20200921132611.1700350-1-elver@google.com The third version of KFENCE patchset is uploaded. Machine lockups on extreme memory pressure (Shakeel Butt) https://lkml.kernel.org/r/CALvZod4FWLsV9byrKQojeus7tMDhHjQHFF5J_JpNsyB0HkaERA@mail.gmail.com Shakeel reports a machine lockup due to extreme memory pressure. It’s locked up on memcg’s vmpressure sr_lock, which is irq-unsafe. And the lockup made in IRQ. He proposes a solution simply changing the lock with irq-safe one, but he also admits it’s not a maintainable solution.

LKML News v5.9-rc6

[PATCH v14 0/5] KASAN-KUnit Integration (David Gow) https://lkml.kernel.org/r/20200915035828.570483-1-davidgow@google.com The 14th patchset for the KASAN and KUnit integration. From the v13, compile warnings in test_kasan_module fixed. [PATCH v2 00/10] KFENCE: A low-overhead sampling-based memory safety error detector (Marco Elver) https://lkml.kernel.org/r/20200915132046.3332537-1-elver@google.com The second version of the KFENCE posted. [PATCH 1/2] fs: Add a filesystem flag for THPs (Matthew Wilcox) https://lkml.kernel.org/r/20200916032717.22917-1-willy@infradead.org This patchset caches whether the backing filesystem support THP or not in an address space flags, so that the page cache can determine if it can send THP to the backing filesystem or not.

LKML News v5.9-rc5

[PATCH RFC 00/10] KFENCE: A low-overhead sampling-based memory safety error detector (Marco Elver) https://lkml.kernel.org/r/20200907134055.2878499-1-elver@google.com This patchset introduces another kernel use-after-free detection infrastructure, called Kernel Electric-Fence (KFENCE). Unlike KASAN, this trades the performance for precision and as a result it incurs near zero performance overhead so that it can be enabled in the production. [PATCH 00/11] Remove assumptions of THP size (Matthew Wilcox) https://lkml.kernel.org/r/20200908195539.25896-1-willy@infradead.org Matthew is working for variable sized THP. For start of the work, this patchset changes the assumption that THP will be PMD size by introducing thp_nr_pages() and let the other VM code to use it.

LKML News v5.9-rc4

[PATCH v9 0/3] introduce memory hinting API for external process (Minchan Kim) https://lkml.kernel.org/r/20200901000633.1920247-1-minchan@kernel.org After the rejection, this patchset posted again. Linus’s opinion about dropping the task_struct to do_madvise and Christian’s comment regarding compat syscall entry has resolved. And, this patchset merged in the ‘-mm’ tree again. Hopefully it will be merged in mainline by 5.10! [PATCH v6 0/3] Anonymous VMA naming patches (Sumit Semwal) https://lkml.kernel.org/r/20200901091901.19779-1-sumit.semwal@linaro.org This is reposting of old patchset which Colin Cross sent before.

LkML New v5.9-rc3

[RFC -V3 0/5] autonuma: Optimize memory placement for memory tiering system (Huang Ying) https://lkml.kernel.org/r/20200825002354.17038-1-ying.huang@intel.com The third RFC of autonuma for help placing of data items on systems utilizing both DRAM and PMEM as different NUMA nodes. Linux 5.9-rc3 https://lkml.kernel.org/r/CAHk-=wiD1yujm_WvWLYL7gZsWMvZyWMg1rfU7dWmnYYc3a7Gsg@mail.gmail.com This week rc release is also quite normal. The ‘fallthought’ comment conversion made fairly evenly distributed changes. Torvalds says it was quite calm. Indeed, only one patchset got my interest this week.

LKML News v5.9-rc2

PROBLEM: Long Workqueue delays. (Jim Baxter) https://lkml.kernel.org/r/71aafe68-7fe0-6b77-ea8e-83edd3f16c8d@mentor.com Jim Baxter encountered 6.5 seconds delay due to long workqueue delay. He want to ask if this is normal or real problem. [RFC PATCH 0/8] memcg: Enable fine-grained per process memory control (Waiman Long) https://lkml.kernel.org/r/20200817140831.30260-1-longman@redhat.com The memory cgroup v2 start reclamation if the amount of the memory of the cgroup exceeds ‘memory.high’ and start OOM killing if if exceeds ‘memory.max’. Problem is, the OOM killer can kill any process.

LKML News v5.9-rc1

mm: sort freelist by rank number (Cho KyongHo) https://lkml.kernel.org/r/1596435031-41837-1-git-send-email-pullip.cho@samsung.com LPDDR5 has severan ranks and there are ranks switching delay. Therefore, it would be better to do accesses in same ranks if possible. This patch sorts freelist by the rank number so that near future allocation can be done in same rank. [RFC-PROTOTYPE 1/1] mm: Add __GFP_FAST_TRY flag (Uladzislau Rezki) https://lkml.kernel.org/r/20200803163029.1997-1-urezki@gmail.com This is the rfc of the patch for the Paul’s last ask.

LKML News v5.8

[PATCH v2 0/7] mm: introduce memfd_secret system call to create “secret” memory areas (Mike Rapoport) https://lkml.kernel.org/r/20200727162935.31714-1-rppt@kernel.org This is the second version of the secretmemfd. In this version, the system call has renamed to ‘memfd_secret’. [RFC PATCH 0/5] madvise MADV_DOEXEC (Anthony Yznaga) https://lkml.kernel.org/r/1595869887-23307-1-git-send-email-anthony.yznaga@oracle.com This patch introduces another madvise hint, MADV_DOEXEC. It preserves an anonymous memory range across exec. [RFC PATCH 0/6] decrease unnecessary gap due to pmem kmem alignment (Jia He) https://lkml.

LKML News v5.8-rc7

[PATCH 0/6] mm: introduce secretmemfd system call to create “secret” memory areas (Mike Rapoport) https://lkml.kernel.org/r/20200720092435.17469-1-rppt@kernel.org This patchset make another special kind of file for secret memory areas. The file can be opened using secretmemfd(). mmap() of the file creates the secret memory mapping. The mapped pages are marked as not present in the direct map and will have desired protection bits (e..g, uncached). [PATCH v7 0/6] workingset protection/detection on the anonymous LRU list (js1304@gmail.