- 12 Feb, 2014 20 commits
-
-
Ben Hutchings authored
These DMA descriptor types will only be used by the userland networking stack. Signed-off-by:
Ben Hutchings <bhutchings@solarflare.com> Signed-off-by:
Shradha Shah <sshah@solarflare.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Ben Hutchings authored
If CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS is defined then NET_IP_ALIGN will be defined as 0, so this macro is redundant. Signed-off-by:
-
Ben Hutchings authored
There is substantial latency in generation and handling of PPS events from the NIC, which we have to correct for before passing a host timestamp to the PPS subsystem. We compare clocks with the MC, giving us two offsets to subtract from the timestamp generated by pps_get_ts(): (a) Time from the last good sync (where we got host and NIC timestamps for nearly the same instant) to the time we called pps_get_ts() (b) Time from NIC top of second to the last good sync We currently calculate (a) + (b) in a quite confusing way. Instead, calculate (a) completely, then add (b) to it. Signed-off-by: -
Ben Hutchings authored
Signed-off-by:
-
Laurence Evans authored
Use conventional net_ratelimit() instead. Signed-off-by:
-
David S. Miller authored
Ben Hutchings says: ==================== ethtool: Improve documentation This series converts most comments in <linux/ethtool.h> into kernel-doc format, and expands them to describe the semantics as well as I understand them. It also fixes some kernel-doc formatting errors. I started on this at Solarflare, but they graciously allowed me to take it away and finish it off. ==================== Signed-off-by: -
Ben Hutchings authored
A colon almost unavoidably starts a new section. The script should be changed to provide a way to avoid this, but for now reword the comments to avoid using colons. Signed-off-by:
Ben Hutchings <ben@decadent.org.uk> Signed-off-by:
-
Ben Hutchings authored
The kernel-doc script does not tolerate the macro definition in between. Signed-off-by:
-
Ben Hutchings authored
Various ethtool command structures are declared with zero-length array at the end which are intended to be variable-length in userland (relying on lack of compiler bounds checking). However, in the kernel the structure and array are always allocated and passed to driver operations separately. Make that explicit. Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Convert the inline comments to kernel-doc format. Explicitly specify that non-zero autoneg is an error if link autonegotiation is disabled. Specify that pause capabilities should be advertised dependent on link autonegotiation, not the autoneg flag here. There is no way to opt-out of pause frame autonegotiation, and this improves behaviour when the link partner is configured to follow pause frame autonegotiation and our interface is not. Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Signed-off-by:
-
Ben Hutchings authored
Replace the inline comments (and some others below) with a full explanation of the semantics, in kernel-doc format. Specify which strings may be empty. Document the relationship with other commands. Replace the 'deprecation' of some fields with a proper explanation of the conversion to generalised string sets, as userland programs may not be able to assume that ETHTOOL_GSSET_INFO is available. Signed-off-by:
-
Ben Hutchings authored
struct ethtool_cmd has very limited documentation; it contains several obscure or obsolete fields and several with non-obvious interpretation. Replace the inline comments (and some others below) with a full explanation of the semantics as well as I understand them, in kernel-doc format. Formally deprecate some fields that seem to be of historical use only. Extend the comment about 32/64-bit compatibility to cover all ethtool structures. Signed-off-by:
-
- 11 Feb, 2014 20 commits
-
-
git://git.monstr.eu/linux-2.6-microblazeLinus Torvalds authored
Pull microblaze fixes from Michal Simek: - Fix two compilation issues - HZ, readq/writeq - Fix stack protection support * tag 'microblaze-3.14-rc3' of git://git.monstr.eu/linux-2.6-microblaze: microblaze: Fix a typo when disabling stack protection microblaze: Define readq and writeq IO helper function microblaze: Fix missing HZ macro
-
git://git.kernel.org/pub/scm/linux/kernel/git/s390/linuxLinus Torvalds authored
Pull s390 bugfixes from Martin Schwidefsky: "A collection a bug fixes. Most of them are minor but two of them are more severe. The linkage stack bug can be used by user space to force an oops, with panic_on_oops this is a denial-of-service. And the dump memory detection issue can cause incomplete memory dumps" * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux: s390/cio: improve cio_commit_config s390: fix kernel crash due to linkage stack instructions s390/dump: Fix dump memory detection s390/appldata: restore missing init_virt_timer() s390/qdio: correct program-controlled interruption checking s390/qdio: for_each macro correctness
-
git://git.kernel.org/pub/scm/linux/kernel/git/davem/netLinus Torvalds authored
Pull networking updates from David Miller: 1) Fix flexcan build on big endian, from Arnd Bergmann 2) Correctly attach cpsw to GPIO bitbang MDIO drive, from Stefan Roese 3) udp_add_offload has to use GFP_ATOMIC since it can be invoked from non-sleepable contexts. From Or Gerlitz 4) vxlan_gro_receive() does not iterate over all possible flows properly, fix also from Or Gerlitz 5) CAN core doesn't use a proper SKB destructor when it hooks up sockets to SKBs. Fix from Oliver Hartkopp 6) ip_tunnel_xmit() can use an uninitialized route pointer, fix from Eric Dumazet 7) Fix address family assignment in IPVS, from Michal Kubecek 8) Fix ath9k build on ARM, from Sujith Manoharan 9) Make sure fail_over_mac only applies for the correct bonding modes, from Ding Tianhong 10) The udp offload code doesn't use RCU correctly, from Shlomo Pongratz 11) Handle gigabit features properly in generic PHY code, from Florian Fainelli 12) Don't blindly invoke link operations in rtnl_link_get_slave_info_data_size, they are optional. Fix from Fernando Luis Vazquez Cao 13) Add USB IDs for Netgear Aircard 340U, from Bjørn Mork 14) Handle netlink packet padding properly in openvswitch, from Thomas Graf 15) Fix oops when deleting chains in nf_tables, from Patrick McHardy 16) Fix RX stalls in xen-netback driver, from Zoltan Kiss 17) Fix deadlock in mac80211 stack, from Emmanuel Grumbach 18) inet_nlmsg_size() forgets to consider ifa_cacheinfo, fix from Geert Uytterhoeven 19) tg3_change_mtu() can deadlock, fix from Nithin Sujir 20) Fix regression in setting SCTP local source addresses on accepted sockets, caused by some generic ipv6 socket changes. Fix from Matija Glavinic Pecotic 21) IPPROTO_* must be pure defines, otherwise module aliases don't get constructed properly. Fix from Jan Moskyto 22) IPV6 netconsole setup doesn't work properly unless an explicit source address is specified, fix from Sabrina Dubroca 23) Use __GFP_NORETRY for high order skb page allocations in sock_alloc_send_pskb and skb_page_frag_refill. From Eric Dumazet 24) Fix a regression added in netconsole over bridging, from Cong Wang 25) TCP uses an artificial offset of 1ms for SRTT, but this doesn't jive well with TCP pacing which needs the SRTT to be accurate. Fix from Eric Dumazet 26) Several cases of missing header file includes from Rashika Kheria 27) Add ZTE MF667 device ID to qmi_wwan driver, from Raymond Wanyoike 28) TCP Small Queues doesn't handle nonagle properly in some corner cases, fix from Eric Dumazet 29) Remove extraneous read_unlock in bond_enslave, whoops. From Ding Tianhong 30) Fix 9p trans_virtio handling of vmalloc buffers, from Richard Yao * git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (136 commits) 6lowpan: fix lockdep splats alx: add missing stats_lock spinlock init 9p/trans_virtio.c: Fix broken zero-copy on vmalloc() buffers bonding: remove unwanted bond lock for enslave processing USB2NET : SR9800 : One chip USB2.0 USB2NET SR9800 Device Driver Support tcp: tsq: fix nonagle handling bridge: Prevent possible race condition in br_fdb_change_mac_address bridge: Properly check if local fdb entry can be deleted when deleting vlan bridge: Properly check if local fdb entry can be deleted in br_fdb_delete_by_port bridge: Properly check if local fdb entry can be deleted in br_fdb_change_mac_address bridge: Fix the way to check if a local fdb entry can be deleted bridge: Change local fdb entries whenever mac address of bridge device changes bridge: Fix the way to find old local fdb entries in br_fdb_change_mac_address bridge: Fix the way to insert new local fdb entries in br_fdb_changeaddr bridge: Fix the way to find old local fdb entries in br_fdb_changeaddr tcp: correct code comment stating 3 min timeout for FIN_WAIT2, we only do 1 min net: vxge: Remove unused device pointer net: qmi_wwan: add ZTE MF667 3c59x: Remove unused pointer in vortex_eisa_cleanup() net: fix 'ip rule' iif/oif device rename ... -
Eric Dumazet authored
When a device ndo_start_xmit() calls again dev_queue_xmit(), lockdep can complain because dev_queue_xmit() is re-entered and the spinlocks protecting tx queues share a common lockdep class. Same issue was fixed for bonding/l2tp/ppp in commits 0daa2303 ("[PATCH] bonding: lockdep annotation") 49ee4920 ("bonding: set qdisc_tx_busylock to avoid LOCKDEP splat") 23d3b8bf ("net: qdisc busylock needs lockdep annotations ") 303c07db ("ppp: set qdisc_tx_busylock to avoid LOCKDEP splat ") Reported-by:
Alexander Aring <alex.aring@gmail.com> Signed-off-by:
Eric Dumazet <edumazet@google.com> Tested-by:
-
John Greene authored
Trivial fix for init time stack trace occuring in alx_get_stats64 upon start up. Should have been part of commit adding the spinlock: f1b6b106 alx: add alx_get_stats64 operation Signed-off-by:
John Greene <jogreene@redhat.com> Signed-off-by:
-
Richard Yao authored
The 9p-virtio transport does zero copy on things larger than 1024 bytes in size. It accomplishes this by returning the physical addresses of pages to the virtio-pci device. At present, the translation is usually a bit shift. That approach produces an invalid page address when we read/write to vmalloc buffers, such as those used for Linux kernel modules. Any attempt to load a Linux kernel module from 9p-virtio produces the following stack. [<ffffffff814878ce>] p9_virtio_zc_request+0x45e/0x510 [<ffffffff814814ed>] p9_client_zc_rpc.constprop.16+0xfd/0x4f0 [<ffffffff814839dd>] p9_client_read+0x15d/0x240 [<ffffffff811c8440>] v9fs_fid_readn+0x50/0xa0 [<ffffffff811c84a0>] v9fs_file_readn+0x10/0x20 [<ffffffff811c84e7>] v9fs_file_read+0x37/0x70 [<ffffffff8114e3fb>] vfs_read+0x9b/0x160 [<ffffffff81153571>] kernel_read+0x41/0x60 [<ffffffff810c83ab>] copy_module_from_fd.isra.34+0xfb/0x180 Subsequently, QEMU will die printing: qemu-system-x86_64: virtio: trying to map MMIO memory This patch enables 9p-virtio to correctly handle this case. This not only enables us to load Linux kernel modules off virtfs, but also enables ZFS file-based vdevs on virtfs to be used without killing QEMU. Special thanks to both Avi Kivity and Alexander Graf for their interpretation of QEMU backtraces. Without their guidence, tracking down this bug would have taken much longer. Also, special thanks to Linus Torvalds for his insightful explanation of why this should use is_vmalloc_addr() instead of is_vmalloc_or_module_addr(): https://lkml.org/lkml/2014/2/8/272Signed-off-by:
Richard Yao <ryao@gentoo.org> Signed-off-by:
-
dingtianhong authored
The bond enslave processing don't hold bond->lock anymore, so release an unlocked rw lock will cause warning message, remove the unwanted read_unlock(&bond->lock). Cc: Jay Vosburgh <fubar@us.ibm.com> Cc: Veaceslav Falico <vfalico@redhat.com> Cc: Andy Gospodarek <andy@greyhouse.net> Signed-off-by:
Ding Tianhong <dingtianhong@huawei.com> Acked-by:
Veaceslav Falico <vfalico@redhat.com> Signed-off-by:
-
Liu Junliang authored
Signed-off-by:
Liu Junliang <liujunliang_ljl@163.com> Signed-off-by:
-
Linus Torvalds authored
Merge misc fixes from Andrew Morton: "A bunch of fixes" * emailed patches fron Andrew Morton <akpm@linux-foundation.org>: ocfs2: check existence of old dentry in ocfs2_link() ocfs2: update inode size after zeroing the hole ocfs2: fix issue that ocfs2_setattr() does not deal with new_i_size==i_size mm/memory-failure.c: move refcount only in !MF_COUNT_INCREASED smp.h: fix x86+cpu.c sparse warnings about arch nonboot CPU calls mm: fix page leak at nfs_symlink() slub: do not assert not having lock in removing freed partial gitignore: add all.config ocfs2: fix ocfs2_sync_file() if filesystem is readonly drivers/edac/edac_mc_sysfs.c: poll timeout cannot be zero fs/file.c:fdtable: avoid triggering OOMs from alloc_fdmem xen: properly account for _PAGE_NUMA during xen pte translations mm/slub.c: list_lock may not be held in some circumstances drivers/md/bcache/extents.c: use %zi to format size_t vmcore: prevent PT_NOTE p_memsz overflow during header update drivers/message/i2o/i2o_config.c: fix deadlock in compat_ioctl(I2OGETIOPS) Documentation/: update 00-INDEX files checkpatch: fix detection of git repository get_maintainer: fix detection of git repository drivers/misc/sgi-gru/grukdump.c: unlocking should be conditional in gru_dump_context()
-
Xue jiufei authored
System call linkat first calls user_path_at(), check the existence of old dentry, and then calls vfs_link()->ocfs2_link() to do the actual work. There may exist a race when Node A create a hard link for file while node B rm it. Node A Node B user_path_at() ->ocfs2_lookup(), find old dentry exist rm file, add inode say inodeA to orphan_dir call ocfs2_link(),create a hard link for inodeA. rm the link, add inodeA to orphan_dir again When orphan_scan work start, it calls ocfs2_queue_orphans() to do the main work. It first tranverses entrys in orphan_dir, linking all inodes in this orphan_dir to a list look like this: inodeA->inodeB->...->inodeA When tranvering this list, it will fall into loop, calling iput() again and again. And finally trigger BUG_ON(inode->i_state & I_CLEAR). Signed-off-by:joyce <xuejiufei@huawei.com> Reviewed-by:
Mark Fasheh <mfasheh@suse.com> Cc: Joel Becker <jlbec@evilplan.org> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Junxiao Bi authored
fs-writeback will release the dirty pages without page lock whose offset are over inode size, the release happens at block_write_full_page_endio(). If not update, dirty pages in file holes may be released before flushed to the disk, then file holes will contain some non-zero data, this will cause sparse file md5sum error. To reproduce the bug, find a big sparse file with many holes, like vm image file, its actual size should be bigger than available mem size to make writeback work more frequently, tar it with -S option, then keep untar it and check its md5sum again and again until you get a wrong md5sum. Signed-off-by:
Junxiao Bi <junxiao.bi@oracle.com> Cc: Younger Liu <younger.liu@huawei.com> Reviewed-by:
Mark Fasheh <mfasheh@suse.de> Cc: Joel Becker <jlbec@evilplan.org> Signed-off-by:
-
Younger Liu authored
The issue scenario is as following: - Create a small file and fallocate a large disk space for a file with FALLOC_FL_KEEP_SIZE option. - ftruncate the file back to the original size again. but the disk free space is not changed back. This is a real bug that be fixed in this patch. In order to solve the issue above, we modified ocfs2_setattr(), if attr->ia_size != i_size_read(inode), It calls ocfs2_truncate_file(), and truncate disk space to attr->ia_size. Signed-off-by:
Younger Liu <younger.liu@huawei.com> Reviewed-by:
Jie Liu <jeff.liu@oracle.com> Tested-by:
Jensen <shencanquan@huawei.com> Signed-off-by:
-
Naoya Horiguchi authored
mce-test detected a test failure when injecting error to a thp tail page. This is because we take page refcount of the tail page in madvise_hwpoison() while the fix in commit a3e0f9e4 ("mm/memory-failure.c: transfer page count from head page to tail page after split thp") assumes that we always take refcount on the head page. When a real memory error happens we take refcount on the head page where memory_failure() is called without MF_COUNT_INCREASED set, so it seems to me that testing memory error on thp tail page using madvise makes little sense. This patch cancels moving refcount in !MF_COUNT_INCREASED for valid testing. [akpm@linux-foundation.org: s/&&/&/] Signed-off-by:
Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: Andi Kleen <andi@firstfloor.org> Cc: Wanpeng Li <liwanp@linux.vnet.ibm.com> Cc: Chen Gong <gong.chen@linux.intel.com> Cc: <stable@vger.kernel.org> [3.9+: a3e0f9e4] Signed-off-by:
-
Paul Gortmaker authored
Use what we already do for arch_disable_smp_support() to fix these: arch/x86/kernel/smpboot.c:1155:6: warning: symbol 'arch_enable_nonboot_cpus_begin' was not declared. Should it be static? arch/x86/kernel/smpboot.c:1160:6: warning: symbol 'arch_enable_nonboot_cpus_end' was not declared. Should it be static? kernel/cpu.c:512:13: warning: symbol 'arch_enable_nonboot_cpus_begin' was not declared. Should it be static? kernel/cpu.c:516:13: warning: symbol 'arch_enable_nonboot_cpus_end' was not declared. Should it be static? Signed-off-by:
Paul Gortmaker <paul.gortmaker@windriver.com> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Signed-off-by:
-
Rafael Aquini authored
Changes in commit a0b8cab3 ("mm: remove lru parameter from __pagevec_lru_add and remove parts of pagevec API") have introduced a call to add_to_page_cache_lru() which causes a leak in nfs_symlink() as now the page gets an extra refcount that is not dropped. Jan Stancek observed and reported the leak effect while running test8 from Connectathon Testsuite. After several iterations over the test case, which creates several symlinks on a NFS mountpoint, the test system was quickly getting into an out-of-memory scenario. This patch fixes the page leak by dropping that extra refcount add_to_page_cache_lru() is grabbing. Signed-off-by:
Jan Stancek <jstancek@redhat.com> Signed-off-by:
Rafael Aquini <aquini@redhat.com> Acked-by:
Mel Gorman <mgorman@suse.de> Acked-by:
Rik van Riel <riel@redhat.com> Cc: Jeff Layton <jlayton@redhat.com> Cc: Trond Myklebust <trond.myklebust@primarydata.com> Cc: <stable@vger.kernel.org> [3.11.x+] Signed-off-by:
-
Steven Rostedt authored
Vladimir reported the following issue: Commit c65c1877 ("slub: use lockdep_assert_held") requires remove_partial() to be called with n->list_lock held, but free_partial() called from kmem_cache_close() on cache destruction does not follow this rule, leading to a warning: WARNING: CPU: 0 PID: 2787 at mm/slub.c:1536 __kmem_cache_shutdown+0x1b2/0x1f0() Modules linked in: CPU: 0 PID: 2787 Comm: modprobe Tainted: G W 3.14.0-rc1-mm1+ #1 Hardware name: 0000000000000600 ffff88003ae1dde8 ffffffff816d9583 0000000000000600 0000000000000000 ffff88003ae1de28 ffffffff8107c107 0000000000000000 ffff880037ab2b00 ffff88007c240d30 ffffea0001ee5280 ffffea0001ee52a0 Call Trace: __kmem_cache_shutdown+0x1b2/0x1f0 kmem_cache_destroy+0x43/0xf0 xfs_destroy_zones+0x103/0x110 [xfs] exit_xfs_fs+0x38/0x4e4 [xfs] SyS_delete_module+0x19a/0x1f0 system_call_fastpath+0x16/0x1b His solution was to add a spinlock in order to quiet lockdep. Although there would be no contention to adding the lock, that lock also requires disabling of interrupts which will have a larger impact on the system. Instead of adding a spinlock to a location where it is not needed for lockdep, make a __remove_partial() function that does not test if the list_lock is held, as no one should have it due to it being freed. Also added a __add_partial() function that does not do the lock validation either, as it is not needed for the creation of the cache. Signed-off-by:
Steven Rostedt <rostedt@goodmis.org> Reported-by:
Vladimir Davydov <vdavydov@parallels.com> Suggested-by:
David Rientjes <rientjes@google.com> Acked-by:
Christoph Lameter <cl@linux.com> Cc: Pekka Enberg <penberg@kernel.org> Signed-off-by:
-
Borislav Petkov authored
This is used by kbuild to load preset Kconfig options. We need to ignore it, otherwise git clean kills it. Signed-off-by:
Borislav Petkov <bp@suse.de> Cc: Michal Marek <mmarek@suse.cz> Signed-off-by:
-
Younger Liu authored
If filesystem is readonly, there is no need to flush drive's caches or force any uncommitted transactions. [akpm@linux-foundation.org: return -EROFS, not 0] Signed-off-by:
Younger Liu <younger.liucn@gmail.com> Cc: Joel Becker <jlbec@evilplan.org> Cc: Mark Fasheh <mfasheh@suse.com> Signed-off-by:
-
Prarit Bhargava authored
If you do echo 0 > /sys/module/edac_core/parameters/edac_mc_poll_msec the following stack trace is output because the edac module is not designed to poll with a timeout of zero. WARNING: CPU: 12 PID: 0 at lib/list_debug.c:33 __list_add+0xac/0xc0() list_add corruption. prev->next should be next (ffff8808291dd1b8), but was (null). (prev=ffff8808286fe3f8). Modules linked in: sg nfsv3 rpcsec_gss_krb5 nfsv4 dns_resolver nfs fscache cfg80211 rfkill x86_pkg_temp_thermal coretemp kvm_intel kvm ixgbe e1000e crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd iTCO_wdt ptp sb_edac iTCO_vendor_support pps_core mdio ipmi_devintf edac_core ioatdma microcode shpchp lpc_ich pcspkr i2c_i801 dca mfd_core ipmi_si wmi ipmi_msghandler nfsd auth_rpcgss nfs_acl lockd sunrpc xfs libcrc32c sd_mod sr_mod cdrom crc_t10dif crct10dif_common mgag200 syscopyarea sysfillrect sysimgblt isci i2c_algo_bit drm_kms_helper ttm drm libsas ahci libahci scsi_transport_sas libata i2c_core dm_mirror dm_region_hash dm_log dm_mod CPU: 12 PID: 0 Comm: swapper/12 Not tainted 3.13.0+ #1 Hardware name: Intel Corporation LH Pass ........../SVRBD-ROW_T, BIOS SE5C600.86B.01.08.0003.022620131521 02/26/2013 Call Trace: <IRQ> __list_add+0xac/0xc0 __internal_add_timer+0xab/0x130 internal_add_timer+0x17/0x40 mod_timer_pinned+0xca/0x170 intel_pstate_timer_func+0x28a/0x380 call_timer_fn+0x36/0x100 run_timer_softirq+0x1ff/0x2f0 __do_softirq+0xf5/0x2e0 irq_exit+0x10d/0x120 smp_apic_timer_interrupt+0x45/0x60 apic_timer_interrupt+0x6d/0x80 <EOI> cpuidle_idle_call+0xb9/0x1f0 arch_cpu_idle+0xe/0x30 cpu_startup_entry+0x9e/0x240 start_secondary+0x1e4/0x290 kernel BUG at kernel/timer.c:1084! invalid opcode: 0000 [#1] SMP Modules linked in: sg nfsv3 rpcsec_gss_krb5 nfsv4 dns_resolver nfs fscache cfg80211 rfkill x86_pkg_temp_thermal coretemp kvm_intel kvm ixgbe e1000e crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd iTCO_wdt ptp sb_edac iTCO_vendor_support pps_core mdio ipmi_devintf edac_core ioatdma microcode shpchp lpc_ich pcspkr i2c_i801 dca mfd_core ipmi_si wmi ipmi_msghandler nfsd auth_rpcgss nfs_acl lockd sunrpc xfs libcrc32c sd_mod sr_mod cdrom crc_t10dif crct10dif_common mgag200 syscopyarea sysfillrect sysimgblt isci i2c_algo_bit drm_kms_helper ttm drm libsas ahci libahci scsi_transport_sas libata i2c_core dm_mirror dm_region_hash dm_log dm_mod CPU: 12 PID: 0 Comm: swapper/12 Tainted: G W 3.13.0+ #1 Hardware name: Intel Corporation LH Pass ........../SVRBD-ROW_T, BIOS SE5C600.86B.01.08.0003.022620131521 02/26/2013 Call Trace: <IRQ> run_timer_softirq+0x245/0x2f0 __do_softirq+0xf5/0x2e0 irq_exit+0x10d/0x120 smp_apic_timer_interrupt+0x45/0x60 apic_timer_interrupt+0x6d/0x80 <EOI> cpuidle_idle_call+0xb9/0x1f0 arch_cpu_idle+0xe/0x30 cpu_startup_entry+0x9e/0x240 start_secondary+0x1e4/0x290 RIP cascade+0x93/0xa0 WARNING: CPU: 36 PID: 1154 at kernel/workqueue.c:1461 __queue_delayed_work+0xed/0x1a0() Modules linked in: sg nfsv3 rpcsec_gss_krb5 nfsv4 dns_resolver nfs fscache cfg80211 rfkill x86_pkg_temp_thermal coretemp kvm_intel kvm ixgbe e1000e crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel aesni_intel lrw gf128mul glue_helper ablk_helper cryptd iTCO_wdt ptp sb_edac iTCO_vendor_support pps_core mdio ipmi_devintf edac_core ioatdma microcode shpchp lpc_ich pcspkr i2c_i801 dca mfd_core ipmi_si wmi ipmi_msghandler nfsd auth_rpcgss nfs_acl lockd sunrpc xfs libcrc32c sd_mod sr_mod cdrom crc_t10dif crct10dif_common mgag200 syscopyarea sysfillrect sysimgblt isci i2c_algo_bit drm_kms_helper ttm drm libsas ahci libahci scsi_transport_sas libata i2c_core dm_mirror dm_region_hash dm_log dm_mod CPU: 36 PID: 1154 Comm: kworker/u481:3 Tainted: G W 3.13.0+ #1 Hardware name: Intel Corporation LH Pass ........../SVRBD-ROW_T, BIOS SE5C600.86B.01.08.0003.022620131521 02/26/2013 Workqueue: edac-poller edac_mc_workq_function [edac_core] Call Trace: dump_stack+0x45/0x56 warn_slowpath_common+0x7d/0xa0 warn_slowpath_null+0x1a/0x20 __queue_delayed_work+0xed/0x1a0 queue_delayed_work_on+0x27/0x50 edac_mc_workq_function+0x72/0xa0 [edac_core] process_one_work+0x17b/0x460 worker_thread+0x11b/0x400 kthread+0xd2/0xf0 ret_from_fork+0x7c/0xb0 This patch adds a range check in the edac_mc_poll_msec code to check for 0. Signed-off-by:Prarit Bhargava <prarit@redhat.com> Cc: Doug Thompson <dougthompson@xmission.com> Cc: Mauro Carvalho Chehab <mchehab@infradead.org> Signed-off-by:
-
Eric W. Biederman authored
Recently due to a spike in connections per second memcached on 3 separate boxes triggered the OOM killer from accept. At the time the OOM killer was triggered there was 4GB out of 36GB free in zone 1. The problem was that alloc_fdtable was allocating an order 3 page (32KiB) to hold a bitmap, and there was sufficient fragmentation that the largest page available was 8KiB. I find the logic that PAGE_ALLOC_COSTLY_ORDER can't fail pretty dubious but I do agree that order 3 allocations are very likely to succeed. There are always pathologies where order > 0 allocations can fail when there are copious amounts of free memory available. Using the pigeon hole principle it is easy to show that it requires 1 page more than 50% of the pages being free to guarantee an order 1 (8KiB) allocation will succeed, 1 page more than 75% of the pages being free to guarantee an order 2 (16KiB) allocation will succeed and 1 page more than 87.5% of the pages being free to guarantee an order 3 allocate will succeed. A server churning memory with a lot of small requests and replies like memcached is a common case that if anything can will skew the odds against large pages being available. Therefore let's not give external applications a practical way to kill linux server applications, and specify __GFP_NORETRY to the kmalloc in alloc_fdmem. Unless I am misreading the code and by the time the code reaches should_alloc_retry in __alloc_pages_slowpath (where __GFP_NORETRY becomes signification). We have already tried everything reasonable to allocate a page and the only thing left to do is wait. So not waiting and falling back to vmalloc immediately seems like the reasonable thing to do even if there wasn't a chance of triggering the OOM killer. Signed-off-by:
"Eric W. Biederman" <ebiederm@xmission.com> Cc: Eric Dumazet <eric.dumazet@gmail.com> Acked-by:
-
