RHEL5.1 Updated kernel package 2.6.18-53.1.19.el5 Oracle's EL 2.6.18-53.1.19.0.1.el5
RHEL kernel-2.6.18-53.1.19.el5 [RHSA-2008:0233-01] Important: kernel security and bug fix update 2008-05-07
Oracle's EL kernel-2.6.18-53.1.19.0.1.el5 [El-errata] ELSA-2008-0233 Important: Enterprise Linux 5 kernel security and bug fix update 2008-05-08
より
Description of changes: [2.6.18-53.1.19.0.1.el5] - [NET] Add entropy support to e1000 and bnx2 (John Sobecki) [ORA 6045759] - [NET] Fix msi issue with kexec/kdump (Michael Chan) [ORA 6219364] - [MM] Fix alloc_pages_node() static `nid' race made kernel crash (Joe Jin) [ORA 6187457] - [splice] Fix bad unlock_page() in error case (Jens Axboe) [ORA 6263574] - [dio] fix error-path crashes (Linux Torvalds) [ORA 6242289] [2.6.18-53.1.19.el5] - [xen] check num of segments in block backend driver (Bill Burns ) [378281] - [x86_64] update IO-APIC dest field to 8-bit for xAPIC (Dave Anderson ) [442922] - Update: [fs] fix race condition in dnotify (Alexander Viro ) [439758] {CVE-2008-1375} - Update: [xen] ia64: ftp stress test fixes between HVM/Dom0 (Tetsu Yamamoto ) [427400] {CVE-2008-1619} [2.6.18-53.1.18.el5] - Update: [fs] fix race condition in dnotify (Alexander Viro ) [439758] {CVE-2008-1375} [2.6.18-53.1.17.el5] - [fs] fix race condition in dnotify (Alexander Viro ) [439758] {CVE-2008-1375} - [pci] hotplug: PCI Express problems with bad DLLPs (Kei Tokunaga ) [440438] - [nfs] stop sillyrenames and unmounts from racing (Steve Dickson ) [440447] - [x86] clear df flag for signal handlers (Jason Baron ) [437316] {CVE-2008-1367} - [xen] ia64: ftp stress test fixes between HVM/Dom0 (Tetsu Yamamoto ) [427400] {CVE-2008-1619} - [xen] ia64: fix ssm_i emulation barrier and vdso pv (Tetsu Yamamoto ) [427400] {CVE-2008-1619} [2.6.18-53.1.16.el5] - [misc] fix range check in fault handlers with mremap (Vitaly Mayatskikh ) [428970] - [video] neofb: avoid overwriting fb_info fields (Anton Arapov ) [430253] [2.6.18-53.1.15.el5] - [libata] sata_nv: un-blacklist hitachi drives (David Milburn ) [433617] - [libata] sata_nv: may send cmds with duplicate tags (David Milburn ) [433617] - [s390] qdio: output queue stall on FCP and net devs (Hans-Joachim Picht ) [412071] - [xen] ia64: guest has bad network performance (Tetsu Yamamoto ) [433616]
Oracle's ELのみの変更は kernel-2.6.18-53.1.19.0.1.el5 と 2.6.18-53.1.14.0.1.el5の間に違いは無い。
しかし、2.6.18-53.1.14.0.1.el5 と 2.6.18-53.1.13.0.1.el5 は違うし、kernel-2.6.18-53.1.13.0.1.el5と kernel-2.6.18-53.1.6.0.1.el5 も違う。ただし、kernel-2.6.18-53.1.6.0.1.el5 と 2.6.18-53.1.14.0.1.el5 と kernel-2.6.18-53.1.19.0.1.el5 との間で Oracle's ELのみの変更に違いはない。このため、kernel-2.6.18-53.1.19.0.1.el5への変更は、2.6.18-53.1.14.el5 と 2.6.18-53.1.14.0.1.el5の差分と、kernel-2.6.18-53.1.6.el5とkernel-2.6.18-53.1.6.0.1.el5の差分と同じし、kernel-2.6.18-53.el5 への Bug Fixes added by Oracleと同じ。ただし、2.6.18-53.1.13.0.1.el5 での Bug Fixes added by Oracleの一部は、2.6.18-53.1.14.el5 に含まれた。
このことを書いてなかったので、メモ。
RHEL5.1 Updated kernel package kernel-2.6.18-53.1.13.el5 Oracle's EL kernel-2.6.18-53.1.13.0.1.el5 - tmpfile::memoで
Oracle's ELのみの変更はkernel-2.6.18-53.1.6.0.1.el5に比べると増えてない。
と書いたのは間違い
Description of changes: [2.6.18-53.1.13.0.1.el5] - [NET] Add entropy support to e1000 and bnx2 (John Sobecki) [orabug 6045759] - [NET] Fix msi issue with kexec/kdump (Michael Chan) [orabug 6219364] - [MM] Fix alloc_pages_node() static `nid' race made kernel crash (Joe Jin) [orabug 6187457] - [splice] Fix bad unlock_page() in error case (Jens Axboe) [orabug 6263574] - [dio] fix error-path crashes (Linus Torvalds) [orabug 6242289] - [MM] Fix leak in hugepages, regression for shared pagetables patch (Adam Litke) [orabug 6732368]
のうち、
- [MM] Fix leak in hugepages, regression for shared pagetables patch (Adam Litke) [orabug 6732368]
が増えてた。
これがRHEL5.1 Updated kernel package 2.6.18-53.1.14.el5 Oracle's EL 2.6.18-53.1.14.0.1.el5 - tmpfile::memoのときに、2.6.18-53.1.14.el5 に含まれた。
[2.6.18-53.1.14.0.1.el5] - [NET] Add entropy support to e1000 and bnx2 (John Sobecki) [orabug 6045759] - [NET] Fix msi issue with kexec/kdump (Michael Chan) [orabug 6219364] - [MM] Fix alloc_pages_node() static `nid' race made kernel crash (Joe Jin) [orabug 6187457] - [splice] Fix bad unlock_page() in error case (Jens Axboe) [orabug 6263574] - [dio] fix error-path crashes (Linus Torvalds) [orabug 6242289] [2.6.18-53.1.14.el5] - merge from 2.6.18-53.1.13 to 2.6.18-53.1.12 - [nfs] potential file corruption issue when writing (Jeff Layton ) [432078] - [ppc] chrp: fix possible strncmp NULL pointer usage (Vitaly Mayatskikh ) [396821] - [isdn] i4l: fix memory overruns (Vitaly Mayatskikh ) [425171] - [isdn] fix possible isdn_net buffer overflows (Aristeu Rozanski ) [392151] {CVE-2007-6063} - [mm] hugepages: leak due to pagetable page sharing (Larry Woodman ) [431522] - [net] NULL dereference in iwl driver (Vitaly Mayatskikh ) [401421] {CVE-2007-5938} - [misc] Denial of service with wedged processes (Jerome Marchand ) [221403] - [xen] ia64: hvm guest memory range checking (Jarod Wilson ) [408701]
431522 – RHEL 5.1 regression in hugepages due to pagetable sharing patchより
Opened by RHEL Product and Program Management on 2008-02-05 03:55 EST [reply] This bug has been copied from bug #428612 and has been proposed to be backported to 5.1 z-stream (EUS).
そして
428612 – RHEL 5.1 regression in hugepages due to pagetable sharing patchより
Opened by John Sobecki on 2008-01-14 01:45 EST [reply] Description of problem: We (Oracle) have had several customers upgrade to RHEL5.1 or Oracle EL 5.1 and report that hugepages on x86_64 are not properly freed after the database SGA/instance is shutdown. Version-Release number of selected component (if applicable): 2.6.18-53.el5 How reproducible: 100% everytime but you need an SGA > 4GB. Steps to Reproduce: 1. Create an SGA > 4Gb and have hugepages allocated for it 2. Shutdown DB and all allocated hugepages will not be freed 3. Despite ipcs -a showing clean and ps -ef showing no orcl processes, some of the SGA pages remain allocated Actual results: As above. Expected results: All allocated hugepages should be released upon DB shutdown. Additional info: Oracle had build a kernel without patch: linux-2.6-mm-shared-page-table-for-hugetlb-page.patch This was added per changelog entry: - [mm] shared page table for hugetlb page (Larry Woodman ) [222753] Without this patch, hugepages are properly released on DB shutdown.
ということ。