From: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com> To: vgoyal@redhat.com, ebiederm@xmission.com, cpw@sgi.com, kumagai-atsushi@mxc.nes.nec.co.jp, lisa.mitchell@hp.com, heiko.carstens@de.ibm.com, akpm@linux-foundation.org Cc: kexec@lists.infradead.org, linux-kernel@vger.kernel.org Subject: [PATCH v2 00/20] kdump, vmcore: support mmap() on /proc/vmcore Date: Sat, 02 Mar 2013 17:35:48 +0900 [thread overview] Message-ID: <20130302083447.31252.93914.stgit@localhost6.localdomain6> (raw) Currently, read to /proc/vmcore is done by read_oldmem() that uses ioremap/iounmap per a single page. For example, if memory is 1GB, ioremap/iounmap is called (1GB / 4KB)-times, that is, 262144 times. This causes big performance degradation. In particular, the current main user of this mmap() is makedumpfile, which not only reads memory from /proc/vmcore but also does other processing like filtering, compression and IO work. Update of page table and the following TLB flush makes such processing much slow; though I have yet to make patch for makedumpfile and yet to confirm how it's improved. To address the issue, this patch implements mmap() on /proc/vmcore to improve read performance. My simple benchmark shows the improvement from 200 [MiB/sec] to over 50.0 [GiB/sec]. ChangeLog ========= v1 => v2) - Clean up the existing codes: use e_phoff, and remove the assumption on PT_NOTE entries. => See PATCH 01, 02. - Fix potencial bug that ELF haeader size is not included in exported vmcoreinfo size. => See Patch 03. - Divide patch modifying read_vmcore() into two: clean-up and primary code change. => See Patch 9, 10. - Put ELF note segments in page-size boundary on the 1st kernel instead of copying them into the buffer on the 2nd kernel. => See Patch 11, 12, 13, 14, 16. Benchmark ========= No change is seen from the previous patch series. See the previous one from here: https://lkml.org/lkml/2013/2/14/89 TODO ==== - fix makedumpfile to use mmap() on /proc/vmcore and benchmark it to confirm whether we can see enough performance improvement. The idea is described here: http://lists.infradead.org/pipermail/kexec/2013-February/007982.html - fix crash utility and makedumpfile to support NT_VMCORE_PAD note type. Both tools don't distinguish the same note types from different note names, which is not conform to ELF specification; now both reads NT_VMCORE_PAD note type as NT_VMCORE_DEBUGINFO. Test ==== This patch set is composed based on v3.9-rc1. Done on x86-64, x86-32 both with 1GB and over 4GB memory environments. --- HATAYAMA Daisuke (20): vmcore: introduce mmap_vmcore() vmcore: count holes generated by round-up operation for vmcore size vmcore: round-up offset of vmcore object in page-size boundary vmcore: check if vmcore objects satify mmap()'s page-size boundary requirement vmcore: check NT_VMCORE_PAD as a mark indicating the end of ELF note buffer kexec: fill note buffers by NT_VMCORE_PAD notes in page-size boundary elf: introduce NT_VMCORE_PAD type kexec, elf: introduce NT_VMCORE_DEBUGINFO note type kexec: allocate vmcoreinfo note buffer on page-size boundary vmcore: allocate per-cpu crash_notes objects on page-size boundary vmcore: read buffers for vmcore objects copied from old memory vmcore: clean up read_vmcore() vmcore: modify vmcore clean-up function to free buffer on 2nd kernel vmcore: copy non page-size aligned head and tail pages in 2nd kernel vmcore, procfs: introduce a flag to distinguish objects copied in 2nd kernel vmcore: round up buffer size of ELF headers by PAGE_SIZE vmcore: allocate buffer for ELF headers on page-size alignment vmcore, sysfs: export ELF note segment size instead of vmcoreinfo data size vmcore: rearrange program headers without assuming consequtive PT_NOTE entries vmcore: refer to e_phoff member explicitly arch/s390/include/asm/kexec.h | 7 fs/proc/vmcore.c | 577 ++++++++++++++++++++++++++++++++--------- include/linux/kexec.h | 16 + include/linux/proc_fs.h | 8 - include/uapi/linux/elf.h | 5 kernel/kexec.c | 47 ++- kernel/ksysfs.c | 2 7 files changed, 505 insertions(+), 157 deletions(-) -- Thanks. HATAYAMA, Daisuke
WARNING: multiple messages have this Message-ID (diff)
From: HATAYAMA Daisuke <d.hatayama@jp.fujitsu.com> To: vgoyal@redhat.com, ebiederm@xmission.com, cpw@sgi.com, kumagai-atsushi@mxc.nes.nec.co.jp, lisa.mitchell@hp.com, heiko.carstens@de.ibm.com, akpm@linux-foundation.org Cc: kexec@lists.infradead.org, linux-kernel@vger.kernel.org Subject: [PATCH v2 00/20] kdump, vmcore: support mmap() on /proc/vmcore Date: Sat, 02 Mar 2013 17:35:48 +0900 [thread overview] Message-ID: <20130302083447.31252.93914.stgit@localhost6.localdomain6> (raw) Currently, read to /proc/vmcore is done by read_oldmem() that uses ioremap/iounmap per a single page. For example, if memory is 1GB, ioremap/iounmap is called (1GB / 4KB)-times, that is, 262144 times. This causes big performance degradation. In particular, the current main user of this mmap() is makedumpfile, which not only reads memory from /proc/vmcore but also does other processing like filtering, compression and IO work. Update of page table and the following TLB flush makes such processing much slow; though I have yet to make patch for makedumpfile and yet to confirm how it's improved. To address the issue, this patch implements mmap() on /proc/vmcore to improve read performance. My simple benchmark shows the improvement from 200 [MiB/sec] to over 50.0 [GiB/sec]. ChangeLog ========= v1 => v2) - Clean up the existing codes: use e_phoff, and remove the assumption on PT_NOTE entries. => See PATCH 01, 02. - Fix potencial bug that ELF haeader size is not included in exported vmcoreinfo size. => See Patch 03. - Divide patch modifying read_vmcore() into two: clean-up and primary code change. => See Patch 9, 10. - Put ELF note segments in page-size boundary on the 1st kernel instead of copying them into the buffer on the 2nd kernel. => See Patch 11, 12, 13, 14, 16. Benchmark ========= No change is seen from the previous patch series. See the previous one from here: https://lkml.org/lkml/2013/2/14/89 TODO ==== - fix makedumpfile to use mmap() on /proc/vmcore and benchmark it to confirm whether we can see enough performance improvement. The idea is described here: http://lists.infradead.org/pipermail/kexec/2013-February/007982.html - fix crash utility and makedumpfile to support NT_VMCORE_PAD note type. Both tools don't distinguish the same note types from different note names, which is not conform to ELF specification; now both reads NT_VMCORE_PAD note type as NT_VMCORE_DEBUGINFO. Test ==== This patch set is composed based on v3.9-rc1. Done on x86-64, x86-32 both with 1GB and over 4GB memory environments. --- HATAYAMA Daisuke (20): vmcore: introduce mmap_vmcore() vmcore: count holes generated by round-up operation for vmcore size vmcore: round-up offset of vmcore object in page-size boundary vmcore: check if vmcore objects satify mmap()'s page-size boundary requirement vmcore: check NT_VMCORE_PAD as a mark indicating the end of ELF note buffer kexec: fill note buffers by NT_VMCORE_PAD notes in page-size boundary elf: introduce NT_VMCORE_PAD type kexec, elf: introduce NT_VMCORE_DEBUGINFO note type kexec: allocate vmcoreinfo note buffer on page-size boundary vmcore: allocate per-cpu crash_notes objects on page-size boundary vmcore: read buffers for vmcore objects copied from old memory vmcore: clean up read_vmcore() vmcore: modify vmcore clean-up function to free buffer on 2nd kernel vmcore: copy non page-size aligned head and tail pages in 2nd kernel vmcore, procfs: introduce a flag to distinguish objects copied in 2nd kernel vmcore: round up buffer size of ELF headers by PAGE_SIZE vmcore: allocate buffer for ELF headers on page-size alignment vmcore, sysfs: export ELF note segment size instead of vmcoreinfo data size vmcore: rearrange program headers without assuming consequtive PT_NOTE entries vmcore: refer to e_phoff member explicitly arch/s390/include/asm/kexec.h | 7 fs/proc/vmcore.c | 577 ++++++++++++++++++++++++++++++++--------- include/linux/kexec.h | 16 + include/linux/proc_fs.h | 8 - include/uapi/linux/elf.h | 5 kernel/kexec.c | 47 ++- kernel/ksysfs.c | 2 7 files changed, 505 insertions(+), 157 deletions(-) -- Thanks. HATAYAMA, Daisuke _______________________________________________ kexec mailing list kexec@lists.infradead.org http://lists.infradead.org/mailman/listinfo/kexec
next reply other threads:[~2013-03-05 7:04 UTC|newest] Thread overview: 82+ messages / expand[flat|nested] mbox.gz Atom feed top 2013-03-02 8:35 HATAYAMA Daisuke [this message] 2013-03-02 8:35 ` [PATCH v2 00/20] kdump, vmcore: support mmap() on /proc/vmcore HATAYAMA Daisuke 2013-03-02 8:35 ` [PATCH v2 01/20] vmcore: refer to e_phoff member explicitly HATAYAMA Daisuke 2013-03-02 8:35 ` HATAYAMA Daisuke 2013-03-05 7:35 ` Zhang Yanfei 2013-03-05 7:35 ` Zhang Yanfei 2013-03-10 6:46 ` Zhang Yanfei 2013-03-10 6:46 ` Zhang Yanfei 2013-03-11 0:31 ` HATAYAMA Daisuke 2013-03-11 0:31 ` HATAYAMA Daisuke 2013-03-11 17:36 ` Vivek Goyal 2013-03-11 17:36 ` Vivek Goyal 2013-03-02 8:35 ` [PATCH v2 02/20] vmcore: rearrange program headers without assuming consequtive PT_NOTE entries HATAYAMA Daisuke 2013-03-02 8:35 ` HATAYAMA Daisuke 2013-03-05 8:36 ` Zhang Yanfei 2013-03-05 8:36 ` Zhang Yanfei 2013-03-05 9:02 ` HATAYAMA Daisuke 2013-03-05 9:02 ` HATAYAMA Daisuke 2013-03-05 9:35 ` Zhang Yanfei 2013-03-05 9:35 ` Zhang Yanfei 2013-03-02 8:36 ` [PATCH v2 03/20] vmcore, sysfs: export ELF note segment size instead of vmcoreinfo data size HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-05 9:29 ` Zhang Yanfei 2013-03-05 9:29 ` Zhang Yanfei 2013-03-06 0:07 ` HATAYAMA Daisuke 2013-03-06 0:07 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 04/20] vmcore: allocate buffer for ELF headers on page-size alignment HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-06 6:57 ` Zhang Yanfei 2013-03-06 6:57 ` Zhang Yanfei 2013-03-06 9:14 ` HATAYAMA Daisuke 2013-03-06 9:14 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 05/20] vmcore: round up buffer size of ELF headers by PAGE_SIZE HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-06 15:51 ` Yanfei Zhang 2013-03-06 15:51 ` Yanfei Zhang 2013-03-02 8:36 ` [PATCH v2 06/20] vmcore, procfs: introduce a flag to distinguish objects copied in 2nd kernel HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-06 15:55 ` Yanfei Zhang 2013-03-06 15:55 ` Yanfei Zhang 2013-03-02 8:36 ` [PATCH v2 07/20] vmcore: copy non page-size aligned head and tail pages " HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-10 6:16 ` Zhang Yanfei 2013-03-10 6:16 ` Zhang Yanfei 2013-03-11 0:27 ` HATAYAMA Daisuke 2013-03-11 0:27 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 08/20] vmcore: modify vmcore clean-up function to free buffer on " HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 09/20] vmcore: clean up read_vmcore() HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 10/20] vmcore: read buffers for vmcore objects copied from old memory HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 11/20] vmcore: allocate per-cpu crash_notes objects on page-size boundary HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-02 8:36 ` [PATCH v2 12/20] kexec: allocate vmcoreinfo note buffer " HATAYAMA Daisuke 2013-03-02 8:36 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 13/20] kexec, elf: introduce NT_VMCORE_DEBUGINFO note type HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 14/20] elf: introduce NT_VMCORE_PAD type HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 15/20] kexec: fill note buffers by NT_VMCORE_PAD notes in page-size boundary HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-07 10:11 ` Zhang Yanfei 2013-03-07 10:11 ` Zhang Yanfei 2013-03-08 1:55 ` HATAYAMA Daisuke 2013-03-08 1:55 ` HATAYAMA Daisuke 2013-03-08 13:02 ` Yanfei Zhang 2013-03-08 13:02 ` Yanfei Zhang 2013-03-09 3:46 ` HATAYAMA Daisuke 2013-03-09 3:46 ` HATAYAMA Daisuke 2013-03-10 2:33 ` Zhang Yanfei 2013-03-10 2:33 ` Zhang Yanfei 2013-03-02 8:37 ` [PATCH v2 16/20] vmcore: check NT_VMCORE_PAD as a mark indicating the end of ELF note buffer HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 17/20] vmcore: check if vmcore objects satify mmap()'s page-size boundary requirement HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 18/20] vmcore: round-up offset of vmcore object in page-size boundary HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 19/20] vmcore: count holes generated by round-up operation for vmcore size HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke 2013-03-02 8:37 ` [PATCH v2 20/20] vmcore: introduce mmap_vmcore() HATAYAMA Daisuke 2013-03-02 8:37 ` HATAYAMA Daisuke
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20130302083447.31252.93914.stgit@localhost6.localdomain6 \ --to=d.hatayama@jp.fujitsu.com \ --cc=akpm@linux-foundation.org \ --cc=cpw@sgi.com \ --cc=ebiederm@xmission.com \ --cc=heiko.carstens@de.ibm.com \ --cc=kexec@lists.infradead.org \ --cc=kumagai-atsushi@mxc.nes.nec.co.jp \ --cc=linux-kernel@vger.kernel.org \ --cc=lisa.mitchell@hp.com \ --cc=vgoyal@redhat.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.