- 13 Nov, 2013 30 commits
-
-
Lukasz Dorau authored
commit 61e4947c upstream. Since: commit 7ceb17e8 md: Allow devices to be re-added to a read-only array. spares are activated on a read-only array. In case of raid1 and raid10 personalities it causes that not-in-sync devices are marked in-sync without checking if recovery has been finished. If a read-only array is degraded and one of its devices is not in-sync (because the array has been only partially recovered) recovery will be skipped. This patch adds checking if recovery has been finished before marking a device in-sync for raid1 and raid10 personalities. In case of raid5 personality such condition is already present (at raid5.c:6029). Bug was introduced in 3.10 and causes data corruption. Signed-off-by:
Pawel Baldysiak <pawel.baldysiak@intel.com> Signed-off-by:
Lukasz Dorau <lukasz.dorau@intel.com> Signed-off-by:
NeilBrown <neilb@suse.de> Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
Bian Yu authored
commit 905b0297 upstream. When operate harddisk and hit errors, md_set_badblocks is called after scsi_restart_operations which already disabled the irq. but md_set_badblocks will call write_sequnlock_irq and enable irq. so softirq can preempt the current thread and that may cause a deadlock. I think this situation should use write_sequnlock_irqsave/irqrestore instead. I met the situation and the call trace is below: [ 638.919974] BUG: spinlock recursion on CPU#0, scsi_eh_13/1010 [ 638.921923] lock: 0xffff8800d4d51fc8, .magic: dead4ead, .owner: scsi_eh_13/1010, .owner_cpu: 0 [ 638.923890] CPU: 0 PID: 1010 Comm: scsi_eh_13 Not tainted 3.12.0-rc5+ #37 [ 638.925844] Hardware name: To be filled by O.E.M. To be filled by O.E.M./MAHOBAY, BIOS 4.6.5 03/05/2013 [ 638.927816] ffff880037ad4640 ffff880118c03d50 ffffffff8172ff85 0000000000000007 [ 638.929829] ffff8800d4d51fc8 ffff880118c03d70 ffffffff81730030 ffff8800d4d51fc8 [ 638.931848] ffffffff81a72eb0 ffff880118c03d90 ffffffff81730056 ffff8800d4d51fc8 [ 638.933884] Call Trace: [ 638.935867] <IRQ> [<ffffffff8172ff85>] dump_stack+0x55/0x76 [ 638.937878] [<ffffffff81730030>] spin_dump+0x8a/0x8f [ 638.939861] [<ffffffff81730056>] spin_bug+0x21/0x26 [ 638.941836] [<ffffffff81336de4>] do_raw_spin_lock+0xa4/0xc0 [ 638.943801] [<ffffffff8173f036>] _raw_spin_lock+0x66/0x80 [ 638.945747] [<ffffffff814a73ed>] ? scsi_device_unbusy+0x9d/0xd0 [ 638.947672] [<ffffffff8173fb1b>] ? _raw_spin_unlock+0x2b/0x50 [ 638.949595] [<ffffffff814a73ed>] scsi_device_unbusy+0x9d/0xd0 [ 638.951504] [<ffffffff8149ec47>] scsi_finish_command+0x37/0xe0 [ 638.953388] [<ffffffff814a75e8>] scsi_softirq_done+0xa8/0x140 [ 638.955248] [<ffffffff8130e32b>] blk_done_softirq+0x7b/0x90 [ 638.957116] [<ffffffff8104fddd>] __do_softirq+0xfd/0x330 [ 638.958987] [<ffffffff810b964f>] ? __lock_release+0x6f/0x100 [ 638.960861] [<ffffffff8174a5cc>] call_softirq+0x1c/0x30 [ 638.962724] [<ffffffff81004c7d>] do_softirq+0x8d/0xc0 [ 638.964565] [<ffffffff8105024e>] irq_exit+0x10e/0x150 [ 638.966390] [<ffffffff8174ad4a>] smp_apic_timer_interrupt+0x4a/0x60 [ 638.968223] [<ffffffff817499af>] apic_timer_interrupt+0x6f/0x80 [ 638.970079] <EOI> [<ffffffff810b964f>] ? __lock_release+0x6f/0x100 [ 638.971899] [<ffffffff8173fa6a>] ? _raw_spin_unlock_irq+0x3a/0x50 [ 638.973691] [<ffffffff8173fa60>] ? _raw_spin_unlock_irq+0x30/0x50 [ 638.975475] [<ffffffff81562393>] md_set_badblocks+0x1f3/0x4a0 [ 638.977243] [<ffffffff81566e07>] rdev_set_badblocks+0x27/0x80 [ 638.978988] [<ffffffffa00d97bb>] raid5_end_read_request+0x36b/0x4e0 [raid456] [ 638.980723] [<ffffffff811b5a1d>] bio_endio+0x1d/0x40 [ 638.982463] [<ffffffff81304ff3>] req_bio_endio.isra.65+0x83/0xa0 [ 638.984214] [<ffffffff81306b9f>] blk_update_request+0x7f/0x350 [ 638.985967] [<ffffffff81306ea1>] blk_update_bidi_request+0x31/0x90 [ 638.987710] [<ffffffff813085e0>] __blk_end_bidi_request+0x20/0x50 [ 638.989439] [<ffffffff8130862f>] __blk_end_request_all+0x1f/0x30 [ 638.991149] [<ffffffff81308746>] blk_peek_request+0x106/0x250 [ 638.992861] [<ffffffff814a62a9>] ? scsi_kill_request.isra.32+0xe9/0x130 [ 638.994561] [<ffffffff814a633a>] scsi_request_fn+0x4a/0x3d0 [ 638.996251] [<ffffffff813040a7>] __blk_run_queue+0x37/0x50 [ 638.997900] [<ffffffff813045af>] blk_run_queue+0x2f/0x50 [ 638.999553] [<ffffffff814a5750>] scsi_run_queue+0xe0/0x1c0 [ 639.001185] [<ffffffff814a7721>] scsi_run_host_queues+0x21/0x40 [ 639.002798] [<ffffffff814a2e87>] scsi_restart_operations+0x177/0x200 [ 639.004391] [<ffffffff814a4fe9>] scsi_error_handler+0xc9/0xe0 [ 639.005996] [<ffffffff814a4f20>] ? scsi_unjam_host+0xd0/0xd0 [ 639.007600] [<ffffffff81072f6b>] kthread+0xdb/0xe0 [ 639.009205] [<ffffffff81072e90>] ? flush_kthread_worker+0x170/0x170 [ 639.010821] [<ffffffff81748cac>] ret_from_fork+0x7c/0xb0 [ 639.012437] [<ffffffff81072e90>] ? flush_kthread_worker+0x170/0x170 This bug was introduce in commit 2e8ac303 (the first time rdev_set_badblock was call from interrupt context), so this patch is appropriate for 3.5 and subsequent kernels. Signed-off-by:
Bian Yu <bianyu@kedacom.com> Reviewed-by:
Jianpeng Ma <majianpeng@gmail.com> Signed-off-by:
-
Gwendal Grignou authored
commit f13e2201 upstream. libata EH decrements scmd->retries when the command failed for reasons unrelated to the command itself so that, for example, commands aborted due to suspend / resume cycle don't get penalized; however, decrementing scmd->retries isn't enough for ATA passthrough commands. Without this fix, ATA passthrough commands are not resend to the drive, and no error is signalled to the caller because: - allowed retry count is 1 - ata_eh_qc_complete fill the sense data, so result is valid - sense data is filled with untouched ATA registers. Signed-off-by:
Gwendal Grignou <gwendal@google.com> Signed-off-by:
Tejun Heo <tj@kernel.org> Signed-off-by:
-
Shaohua Li authored
commit d47648fc upstream. SCSI discard will damage discard stripe bio setting, eg, some fields are changed. If the stripe is reused very soon, we have wrong bios setting. We remove discard stripe from hash list, so next time the strip will be fully initialized. Suitable for backport to 3.7+. Signed-off-by:
Shaohua Li <shli@fusionio.com> Signed-off-by:
-
Shaohua Li authored
commit 37c61ff3 upstream. SCSI layer will add new payload for discard request. If two bios are merged to one, the second bio has bi_vcnt 1 which is set in raid5. This will confuse SCSI and cause oops. Suitable for backport to 3.7+ Reported-by:
Jes Sorensen <Jes.Sorensen@redhat.com> Signed-off-by:
Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by:
-
Geyslan G. Bem authored
commit 3edc8376 upstream. In 'decrypt_pki_encrypted_session_key' function: Initializes 'payload' pointer and releases it on exit. Signed-off-by:
Geyslan G. Bem <geyslan@gmail.com> Signed-off-by:
Tyler Hicks <tyhicks@canonical.com> Signed-off-by:
-
Aaron Lu authored
commit 10c580e4 upstream. Sujit has found a race condition that would make q->nr_pending unbalanced, it occurs as Sujit explained: " sd_probe_async() -> add_disk() -> disk_add_event() -> schedule(disk_events_workfn) sd_revalidate_disk() blk_pm_runtime_init() return; Let's say the disk_events_workfn() calls sd_check_events() which tries to send test_unit_ready() and because of sd_revalidate_disk() trying to send another commands the test_unit_ready() might be re-queued as the tagged command queuing is disabled. So the race condition is - Thread 1 | Thread 2 sd_revalidate_disk() | sd_check_events() ...nr_pending = 0 as q->dev = NULL| scsi_queue_insert() blk_runtime_pm_init() | blk_pm_requeue_request() -> | nr_pending = -1 since | q->dev != NULL " The problem is, the test_unit_ready request doesn't get counted the first time it is queued, so the later decrement of q->nr_pending in blk_pm_requeue_request makes it unbalanced. Fix this by calling blk_pm_runtime_init before add_disk so that all requests initiated there will all be counted. Signed-off-by:
Aaron Lu <aaron.lu@intel.com> Reported-and-tested-by:
Sujit Reddy Thumma <sthumma@codeaurora.org> Signed-off-by:
James Bottomley <JBottomley@Parallels.com> Signed-off-by:
-
Marc Kleine-Budde authored
commit d5a7b406 upstream. In patch 0d1862ea can: flexcan: fix flexcan_chip_start() on imx6 the loop in flexcan_chip_start() that iterates over all mailboxes after the soft reset of the CAN core was removed. This loop put all mailboxes (even the ones marked as reserved 1...7) into EMPTY/INACTIVE mode. On mailboxes 8...63, this aborts any pending TX messages. After a cold boot there is random garbage in the mailboxes, which leads to spontaneous transmit of CAN frames during first activation. Further if the interface was disabled with a pending message (usually due to an error condition on the CAN bus), this message is retransmitted after enabling the interface again. This patch fixes the regression by: 1) Limiting the maximum number of used mailboxes to 8, 0...7 are used by the RX FIFO, 8 is used by TX. 2) Marking the TX mailbox as EMPTY/INACTIVE, so that any pending TX of that mailbox is aborted. Cc: Lothar Waßmann <LW@KARO-electronics.de> Signed-off-by:
Marc Kleine-Budde <mkl@pengutronix.de> Signed-off-by:
-
Marc Kleine-Budde authored
commit e3587842 upstream. The current implemetation of of_match_device() relies that the of_device_id table in the driver is sorted from most specific to least specific compatible. Without this patch the mx28 is detected as the less specific p1010. This leads to a p1010 specific workaround is activated on the mx28, which is not needed. Signed-off-by:
-
Marc Kleine-Budde authored
commit 5abbeea5 upstream. In commit: 3078cde7 can: at91_can: add dt support device tree support was added to the at91_can driver. In this commit the mapping of device to driver data was mixed up. This results in the sam9x5 parameters being used for the sam9263 and the workaround for the broken mailbox 0 on the sam9263 not being activated. This patch fixes the broken platform_device_id table. Cc: Ludovic Desroches <ludovic.desroches@atmel.com> Signed-off-by:
-
Dave Kleikamp authored
commit 86609986 upstream. If insert_inode_locked() fails, we shouldn't be calling unlock_new_inode(). Signed-off-by:
Dave Kleikamp <dave.kleikamp@oracle.com> Tested-by:
Michael L. Semon <mlsemon35@gmail.com> Signed-off-by:
-
Emmanuel Grumbach authored
commit 08a5dd38 upstream. Add some new PCI IDs to the table for 6000, 6005 and 6235 series. Signed-off-by:
Emmanuel Grumbach <emmanuel.grumbach@intel.com> Signed-off-by:
Johannes Berg <johannes.berg@intel.com> Signed-off-by:
-
Mark Cave-Ayland authored
commit 9473ca6e upstream. An error in calculating the offset in an skb causes the driver to read essential device info from the wrong locations. The main effect is that automatic gain calculations are nonsense. Signed-off-by:
Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk> Signed-off-by:
Larry Finger <Larry.Finger@lwfinger.net> Signed-off-by:
John W. Linville <linville@tuxdriver.com> Signed-off-by:
-
Amitkumar Karwar authored
commit 453b0c3f upstream. 601216e1 "mwifiex: process RX packets in SDIO IRQ thread directly" introduced a command timeout issue which can be reproduced easily on an AM33xx platform using a test application written by Daniel Mack: https://gist.github.com/zonque/6579314 mwifiex_main_process() is called from both the SDIO handler and the workqueue. In case an interrupt occurs right after the int_status check, but before updating the mwifiex_processing flag, this interrupt gets lost, resulting in a command timeout and consequently a card reset. Let main_proc_lock protect both int_status and mwifiex_processing flag. This fixes the interrupt lost issue. Reported-by:
Sven Neumann <s.neumann@raumfeld.com> Reported-by:
Andreas Fenkart <andreas.fenkart@streamunlimited.com> Tested-by:
Daniel Mack <zonque@gmail.com> Reviewed-by:
Dylan Reid <dgreid@chromium.org> Signed-off-by:
Amitkumar Karwar <akarwar@marvell.com> Signed-off-by:
Bing Zhao <bzhao@marvell.com> Signed-off-by:
Paul Stewart <pstew@chromium.org> Signed-off-by:
-
Bruno Randolf authored
commit f478f33a upstream. Fix kernel warning when using WEXT for configuring ad-hoc mode, e.g. "iwconfig wlan0 essid test channel 1" WARNING: at net/wireless/chan.c:373 cfg80211_chandef_usable+0x50/0x21c [cfg80211]() The warning is caused by an uninitialized variable center_freq1. Signed-off-by:
Bruno Randolf <br1@einfach.org> Signed-off-by:
-
Felix Fietkau authored
commit ec30326e upstream. Otherwise, if queues are full during a scan, tx scheduling does not resume after switching back to the home channel. Signed-off-by:
Felix Fietkau <nbd@openwrt.org> Signed-off-by:
-
Johannes Berg authored
commit d86aa4f8 upstream. If a frame's timestamp is calculated, and the bitrate calculation goes wrong and returns zero, the system will attempt to divide by zero and crash. Catch this case and print the rate information that the driver reported when this happens. Reported-by:
Thomas Lindroth <thomas.lindroth@gmail.com> Signed-off-by:
-
Felix Fietkau authored
commit 0c5b9329 upstream. When clients are idle for too long, hostapd sends nullfunc frames for probing. When those are acked by the client, the idle time needs to be updated. To make this work (and to avoid unnecessary probing), update sta->last_rx whenever an ACK was received for a tx packet. Only do this if the flag IEEE80211_HW_REPORTS_TX_ACK_STATUS is set. Signed-off-by:
-
Felix Fietkau authored
commit 03bb7f42 upstream. This allows calls for clients in AP_VLANs (e.g. for 4-addr) to succeed Signed-off-by:
-
Felix Fietkau authored
commit 6329b8d9 upstream. If an Ad-Hoc node receives packets with the Cell ID or its own MAC address as source address, it hits a WARN_ON in sta_info_insert_check() With many packets, this can massively spam the logs. One way that this can easily happen is through having Cisco APs in the area with rouge AP detection and countermeasures enabled. Such Cisco APs will regularly send fake beacons, disassoc and deauth packets that trigger these warnings. To fix this issue, drop such spoofed packets early in the rx path. Reported-by:
Thomas Huehn <thomas@net.t-labs.tu-berlin.de> Signed-off-by:
-
Emmanuel Grumbach authored
commit a754055a upstream. __ieee80211_scan_completed is called from a worker. This means that the following flow is possible. * driver calls ieee80211_scan_completed * mac80211 cancels the scan (that is already complete) * __ieee80211_scan_completed runs When scan_work will finally run, it will see that the scan hasn't been aborted and might even trigger another scan on another band. This leads to a situation where cfg80211's scan is not done and no further scan can be issued. Fix this by setting a new flag when a HW scan is being cancelled so that no other scan will be triggered. Signed-off-by:
-
Anjana V Kumar authored
commit ea84753c upstream. Both Anjana and Eunki reported a stall in the while_each_thread loop in cgroup_attach_task(). It's because, when we attach a single thread to a cgroup, if the cgroup is exiting or is already in that cgroup, we won't break the loop. If the task is already in the cgroup, the bug can lead to another thread being attached to the cgroup unexpectedly: # echo 5207 > tasks # cat tasks 5207 # echo 5207 > tasks # cat tasks 5207 5215 What's worse, if the task to be attached isn't the leader of the thread group, we might never exit the loop, hence cpu stall. Thanks for Oleg's analysis. This bug was introduced by commit 081aa458 ("cgroup: consolidate cgroup_attach_task() and cgroup_attach_proc()") [ lizf: - fixed the first continue, pointed out by Oleg, - rewrote changelog. ] Reported-by:
Eunki Kim <eunki_kim@samsung.com> Reported-by:
Anjana V Kumar <anjanavk12@gmail.com> Signed-off-by:
Li Zefan <lizefan@huawei.com> Signed-off-by:
-
Kent Overstreet authored
commit d4eddd42 upstream. Signed-off-by:
Kent Overstreet <kmo@daterainc.com> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by:
-
Dirk Brandewie authored
commit 52e0a509 upstream. If the system is suspended while max_perf_pct is less than 100 percent or no_turbo set policy->{min,max} will be set incorrectly with scaled values which turn the scaled values into hard limits. References: https://bugzilla.kernel.org/show_bug.cgi?id=61241Reported-by:
Patrick Bartels <petzicus@googlemail.com> Signed-off-by:
Dirk Brandewie <dirk.j.brandewie@intel.com> Signed-off-by:
Rafael J. Wysocki <rafael.j.wysocki@intel.com> Signed-off-by:
-
Russ Anderson authored
commit dd3c9c4b upstream. The UV3 hub revision ID is different than expected. The first revision was supposed to start at 1 but instead will start at 0. Signed-off-by:
Russ Anderson <rja@sgi.com> Link: http://lkml.kernel.org/r/20131014161733.GA6274@sgi.comSigned-off-by:
Ingo Molnar <mingo@kernel.org> Signed-off-by:
-
Алексей Крамаренко authored
commit e1466ad5 upstream. Custom VID/PID for Z3X Box device, popular tool for cellphone flashing. Signed-off-by:
Alexey E. Kramarenko <alexeyk13@yandex.ru> Signed-off-by:
-
Oliver Neukum authored
commit 614ced91 upstream. The device descriptors are messed up after remote wakeup Signed-off-by:
Oliver Neukum <oneukum@suse.de> Signed-off-by:
-
Oliver Neukum authored
commit 4294bca7 upstream. The device is not responsive when resumed, unless it is reset. Signed-off-by:
-
Fangxiaozhi (Franko) authored
commit d544db29 upstream. Add new supporting declarations to option.c, to support Huawei new devices with new bInterfaceSubClass value. Signed-off-by:
fangxiaozhi <huananhu@huawei.com> Signed-off-by:
-
Oliver Neukum authored
commit 32c37fc3 upstream. Some USB drive enclosures do not correctly report an overflow condition if they hold a drive with a capacity over 2TB and are confronted with a READ_CAPACITY_10. They answer with their capacity modulo 2TB. The generic layer cannot cope with that. It must be told to use READ_CAPACITY_16 from the beginning. Signed-off-by:
-
- 04 Nov, 2013 10 commits
-
-
Greg Kroah-Hartman authored
-
Enrico Mioso authored
commit fd8573f5 upstream. Interface 6 of this device speaks QMI as per tests done by us. Credits go to Antonella for providing the hardware. Signed-off-by:
Enrico Mioso <mrkiko.rs@gmail.com> Signed-off-by:
Antonella Pellizzari <anto.pellizzari83@gmail.com> Tested-by:
Dan Williams <dcbw@redhat.com> Signed-off-by:
-
Greg Kroah-Hartman authored
commit f4c19b8e upstream. This patch adds the device id for the Inovia SEW858 device to the option driver. Reported-by:
Pavel Parkhomenko <ra85551@gmail.com> Tested-by:
-
Diego Elio Pettenò authored
commit c9d09dc7 upstream. Without this change, the USB cable for Freestyle Option and compatible glucometers will not be detected by the driver. Signed-off-by:
Diego Elio Pettenò <flameeyes@flameeyes.eu> Signed-off-by:
-
Roel Kluin authored
commit d969de8d upstream. Due to missing braces on an if statement, in presence of a device_node a port was always assigned -1, regardless of any alias entries in the device tree. Conversely, if device_node was NULL, an unitialized port ended up being used. This patch adds the missing braces, fixing the issues. Signed-off-by:
Roel Kluin <roel.kluin@gmail.com> Acked-by:
Tony Prisk <linux@prisktech.co.nz> Signed-off-by:
-
Johannes Berg authored
commit f5563318 upstream. When parsing an invalid radiotap header, the parser can overrun the buffer that is passed in because it doesn't correctly check 1) the minimum radiotap header size 2) the space for extended bitmaps The first issue doesn't affect any in-kernel user as they all check the minimum size before calling the radiotap function. The second issue could potentially affect the kernel if an skb is passed in that consists only of the radiotap header with a lot of extended bitmaps that extend past the SKB. In that case a read-only buffer overrun by at most 4 bytes is possible. Fix this by adding the appropriate checks to the parser. Reported-by:
Evan Huus <eapache@gmail.com> Signed-off-by:
-
Fengguang Wu authored
commit e3b6c655 upstream. Toralf runs trinity on UML/i386. After some time it hangs and the last message line is BUG: soft lockup - CPU#0 stuck for 22s! [trinity-child0:1521] It's found that pages_dirtied becomes very large. More than 1000000000 pages in this case: period = HZ * pages_dirtied / task_ratelimit; BUG_ON(pages_dirtied > 2000000000); BUG_ON(pages_dirtied > 1000000000); <--------- UML debug printf shows that we got negative pause here: ick: pause : -984 ick: pages_dirtied : 0 ick: task_ratelimit: 0 pause: + if (pause < 0) { + extern int printf(char *, ...); + printf("ick : pause : %li\n", pause); + printf("ick: pages_dirtied : %lu\n", pages_dirtied); + printf("ick: task_ratelimit: %lu\n", task_ratelimit); + BUG_ON(1); + } trace_balance_dirty_pages(bdi, Since pause is bounded by [min_pause, max_pause] where min_pause is also bounded by max_pause. It's suspected and demonstrated that the max_pause calculation goes wrong: ick: pause : -717 ick: min_pause : -177 ick: max_pause : -717 ick: pages_dirtied : 14 ick: task_ratelimit: 0 The problem lies in the two "long = unsigned long" assignments in bdi_max_pause() which might go negative if the highest bit is 1, and the min_t(long, ...) check failed to protect it falling under 0. Fix all of them by using "unsigned long" throughout the function. Signed-off-by:
Fengguang Wu <fengguang.wu@intel.com> Reported-by:
Toralf Förster <toralf.foerster@gmx.de> Tested-by:
Jan Kara <jack@suse.cz> Cc: Richard Weinberger <richard@nod.at> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Signed-off-by:
Andrew Morton <akpm@linux-foundation.org> Signed-off-by:
-
David Henningsson authored
commit ccb04157 upstream. The create_bind_cap_vol_ctl does not create any control indicating that an inverted dmic is present. Therefore, create multiple capture volumes in this scenario, so we always have some indication that the internal mic is inverted. This happens on the Lenovo Ideapad U310 as well as the Lenovo Yoga 13 (both are based on the CX20590 codec), but the fix is generic and could be needed for other codecs/machines too. Thanks to Szymon Acedański for the pointer and a draft patch. BugLink: https://bugs.launchpad.net/bugs/1239392 BugLink: https://bugs.launchpad.net/bugs/1227491Reported-by:
Szymon Acedański <accek@mimuw.edu.pl> Signed-off-by:
David Henningsson <david.henningsson@canonical.com> Signed-off-by:
Takashi Iwai <tiwai@suse.de> Signed-off-by:
-
Takashi Iwai authored
commit ac536a84 upstream. The pcm_usb_stream plugin requires the mremap explicitly for the read buffer, as it expands itself once after reading the required size. But the commit [314e51b9: mm: kill vma flag VM_RESERVED and mm->reserved_vm counter] converted blindly to a combination of VM_DONTEXPAND | VM_DONTDUMP like other normal drivers, and this resulted in the failure of mremap(). For fixing this regression, we need to remove VM_DONTEXPAND for the read-buffer mmap. Reported-and-tested-by:
James Miller <jamesstewartmiller@gmail.com> Signed-off-by:
-
Hugh Dickins authored
commit 750e8165 upstream. Occasionally we hit the BUG_ON(pmd_trans_huge(*pmd)) at the end of __split_huge_page_pmd(): seen when doing madvise(,,MADV_DONTNEED). It's invalid: we don't always have down_write of mmap_sem there: a racing do_huge_pmd_wp_page() might have copied-on-write to another huge page before our split_huge_page() got the anon_vma lock. Forget the BUG_ON, just go back and try again if this happens. Signed-off-by:
Hugh Dickins <hughd@google.com> Acked-by:
Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Andrea Arcangeli <aarcange@redhat.com> Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com> Cc: David Rientjes <rientjes@google.com> Signed-off-by:
-
