- 10 Jun, 2015 7 commits
-
-
Viresh Kumar authored
cpufreq_update_policy() was kept as a separate routine earlier as it was handling migration of sysfs directories, which isn't the case anymore. It is only updating policy->cpu now and is called by a single caller. The WARN_ON() isn't really required anymore, as we are just updating the cpu now, not moving the sysfs directories. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
__cpufreq_remove_dev_finish() is doing two things today: - Restarts the governor if some CPUs from concerned policy are still online. - Frees the policy if all CPUs are offline. The first task of restarting the governor can be moved to __cpufreq_remove_dev_prepare() to restart the governor early. There is no race between _prepare() and _finish() as they would be handling completely different cases. _finish() will only be required if we are going to free the policy and that has nothing to do with restarting the governor. Original-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
cpufreq_policy_put_kobj() is actually part of freeing the policy and can be called from cpufreq_policy_free() directly instead of a separate call. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
policy->kobj is required to be initialized once in the lifetime of a policy. Currently we are initializing it from __cpufreq_add_dev() and that doesn't look to be the best place for doing so as we have to do this on special cases (like: !recover_policy). We can initialize it from a more obvious place cpufreq_policy_alloc() and that will make code look cleaner, specially the error handling part. The error handling part of __cpufreq_add_dev() was doing almost the same thing while recover_policy is true or false. Fix that as well by always calling cpufreq_policy_put_kobj() with an additional parameter to skip notification part of it. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
When we hot-unplug a cpu, we remove its sysfs cpufreq directory and if the outgoing cpu was the owner of policy->kobj earlier then we migrate the sysfs directory to under another online cpu. There are few disadvantages this brings: - Code Complexity - Slower hotplug/suspend/resume - sysfs file permissions are reset after all policy->cpus are offlined - CPUFreq stats history lost after all policy->cpus are offlined - Special management of sysfs stuff during suspend/resume To overcome these, this patch modifies the way sysfs directories are managed: - Select sysfs kobjects owner while initializing policy and don't change it during hotplugs. Track it with kobj_cpu created earlier. - Create symlinks for all related CPUs (can be offline) instead of affected CPUs on policy initialization and remove them only when the policy is freed. - Free policy structure only on the removal of cpufreq-driver and not during hotplug/suspend/resume, detected by checking 'struct subsys_interface *' (Valid only when called from subsys_interface_unregister() while unregistering driver). Apart from this, special care is taken to handle physical hoplug of CPUs as we wouldn't remove sysfs links or remove policies on logical hotplugs. Physical hotplug happens in the following sequence. Hot removal: - CPU is offlined first, ~ 'echo 0 > /sys/devices/system/cpu/cpuX/online' - Then its device is removed along with all sysfs files, cpufreq core notified with cpufreq_remove_dev() callback from subsys-interface.. Hot addition: - First the device along with its sysfs files is added, cpufreq core notified with cpufreq_add_dev() callback from subsys-interface.. - CPU is onlined, ~ 'echo 1 > /sys/devices/system/cpu/cpuX/online' We call the same routines with both hotplug and subsys callbacks, and we sense physical hotplug with cpu_offline() check in subsys callback. We can handle most of the stuff with regular hotplug callback paths and add/remove cpufreq sysfs links or free policy from subsys callbacks. Original-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
Later commits would change the way policies are managed today. Policies wouldn't be freed on cpu hotplug (currently they aren't freed only for suspend), and while the CPU is offline, the sysfs cpufreq files would still be present. User may accidentally try to update the sysfs files in following directory: '/sys/devices/system/cpu/cpuX/cpufreq/'. And that would result in undefined behavior as policy wouldn't be active then. Apart from updating the store() routine, we also update __cpufreq_get() which can call cpufreq_out_of_sync(). The later routine tries to update policy->cur and starts notifying kernel about it. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Doug Smythies authored
During initialization and exit it is possible that the target pstate might not actually be set. Furthermore, the result can be that the driver and the processor are out of synch and, under some conditions, the driver might never send the processor the proper target pstate. This patch adds a bypass or do_checks flag to the call to intel_pstate_set_pstate. If bypass, then specifically bypass clamp checks and the do not send if it is the same as last time check. If do_checks, then, and as before, do the current policy clamp checks, and do not do actual send if the new target is the same as the old. Signed-off-by: Doug Smythies <dsmythies@telus.net> Reported-by: Marien Zwart <marien.zwart@gmail.com> Reported-by: Alex Lochmann <alexander.lochmann@tu-dortmund.de> Reported-by: Piotr Ko?aczkowski <pkolaczk@gmail.com> Reported-by: Clemens Eisserer <linuxhippy@gmail.com> Tested-by: Marien Zwart <marien.zwart@gmail.com> Tested-by: Doug Smythies <dsmythies@telus.net> [ rjw: Dropped pointless symbol definitions, rebased ] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 09 Jun, 2015 1 commit
-
-
Doug Smythies authored
Commit ce717613 (intel_pstate: Turn per cpu printk into pr_debug) turned per cpu printk into pr_debug. However, only half of the change was done, introducing an inconsistency between entry and exit from driver pstate control. This patch changes the exit message to pr_debug also. The various messages are inconsistent with respect to any identifier text that can be used to help isolate the desired information from a huge log. This patch makes a consistent identifier portion of the string. Amends: ce717613 (intel_pstate: Turn per cpu printk into pr_debug) Signed-off-by: Doug Smythies <dsmythies@telus.net> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 22 May, 2015 2 commits
-
-
Saravana Kannan authored
In order to prepare for the next few commits, that will stop migrating sysfs files on cpu hotplug, this patch starts managing sysfs-cpu separately. The behavior is still the same as we are still migrating sysfs files on hotplug, later commits would change that. Signed-off-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Shailendra Verma authored
Signed-off-by: Shailendra Verma <shailendra.capricorn@gmail.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 15 May, 2015 6 commits
-
-
Viresh Kumar authored
Later commits would change the way policies are managed today. Policies wouldn't be freed on cpu hotplug (currently they aren't freed on suspend), and while the CPU is offline, the sysfs cpufreq files would still be present. Because we don't mark policy->governor as NULL, it still contains pointer of the last used governor. And if the governor is removed, while all the CPUs of a policy are hotplugged out, this pointer wouldn't be valid anymore. And if we try to read the 'scaling_governor', etc. from sysfs, it will result in kernel OOPs. To prevent this, mark policy->governor as NULL for all inactive policies while the governor is removed from kernel. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
History of which governor was used last is common to all CPUs within a policy and maintaining it per-cpu isn't the best approach for sure. Apart from wasting memory, this also increases the complexity of managing this data structure as it has to be updated for all CPUs. To make that somewhat simpler, lets store this information in a new field 'last_governor' in struct cpufreq_policy and update it on removal of last cpu of a policy. As a side-effect it also solves an old problem, consider a system with two clusters 0 & 1. And there is one policy per cluster. Cluster 0: CPU0 and 1. Cluster 1: CPU2 and 3. - CPU2 is first brought online, and governor is set to performance (default as cpufreq_cpu_governor wasn't set). - Governor is changed to ondemand. - CPU2 is taken offline and cpufreq_cpu_governor is updated for CPU2. - CPU3 is brought online. - Because cpufreq_cpu_governor wasn't set for CPU3, the default governor performance is picked for CPU3. This patch fixes the bug as we now have a single variable to update for policy. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
We reach here while adding policy for a CPU and enter into the 'if' block only if a policy already exists for the CPU. As cpufreq_cpu_data is set for all policy->related_cpus now, when the policy is first added, we can use that to find the CPU's policy instead of traversing the list of all active policies. Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
We can extract the same information from cpufreq_cpu_data as it is also available for inactive policies now. And so don't need cpufreq_cpu_data_fallback anymore. Also add a WARN_ON() for the case where we try to restore from an active policy. Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
Now that we can check policy->cpus to find if policy is active or not, we don't need to clean cpufreq_cpu_data and delete policy from the list on light weight tear down of policies (like in suspend). To make it consistent and clean, set cpufreq_cpu_data for all related CPUs when the policy is first created and clean it only while it is freed. Also update cpufreq_cpu_get_raw() to check if cpu is part of policy->cpus mask, so that we don't end up getting policies for offline CPUs. In order to make sure that no users of 'policy' are using an inactive policy, use cpufreq_cpu_get_raw() instead of directly accessing cpufreq_cpu_data. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
policy->cpus is cleared unconditionally now on hotplug-out of a CPU and it can be checked to know if a policy is active or not. Create helper routines to iterate over all active/inactive policies, based on policy->cpus field. Replace all instances of for_each_policy() with for_each_active_policy() to make them iterate only for active policies. (We haven't made changes yet to keep inactive policies in the same list, but that will be followed in a later patch). Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 14 May, 2015 1 commit
-
-
Sudeep Holla authored
With the addition of switcher code, there's compile-time dependency on BIG_LITTLE to get arm_big_little driver compiling on ARM64. Since ARM64 will never add support for bL switcher, it's better to remove the dependency so that the driver can be reused on ARM64 platforms. This patch adds stubs to remove BIG_LITTLE dependency in the driver. Signed-off-by: Sudeep Holla <sudeep.holla@arm.com> Acked-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 12 May, 2015 1 commit
-
-
Joe Konno authored
Commit 007bea09 (intel_pstate: Add setting voltage value for baytrail P states.) introduced byt_set_pstate() with the assumption that it would always be run by the CPU whose MSR is to be written by it. It turns out, however, that is not always the case in practice, so modify byt_set_pstate() to enforce the MSR write done by it to always happen on the right CPU. Fixes: 007bea09 (intel_pstate: Add setting voltage value for baytrail P states.) Signed-off-by: Joe Konno <joe.konno@intel.com> Acked-by: Kristen Carlson Accardi <kristen@linux.intel.com> Cc: 3.14+ <stable@vger.kernel.org> # 3.14+ Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 07 May, 2015 5 commits
-
-
Viresh Kumar authored
We clear policy->cpus mask while CPUs are hotplugged out. We do it for all CPUs except the last CPU of the policy. I don't remember what the rationale behind that was, but I couldn't think of anything that will break if we remove this conditional clearing and always clear policy->cpus. The benefit we get out of it is, we can know if a policy is active or not by checking if this field is empty or not. That will be used by later commits. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
There are two cases when we may try to add CPUs we're already handling: - On boot, the first cpu has marked all policy->cpus managed and so we will find policy for all other policy->cpus later on. - When a managed cpu is hotplugged out and later brought back in. Currently, separate paths and checks take care of the two. While the first one is detected by testing cpu against 'policy->cpus', the other one is detected by testing cpu against 'policy->related_cpus'. We can handle them both via a single path and there is no need to do special checking for the first one. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Saravana Kannan <skannan@codeaurora.org> [ rjw: Changelog, comments ] Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
Simply returning here with an error is not enough. It shouldn't be allowed at all to try calling cpufreq_cpu_get() for an invalid CPU. Add a WARN here to make it clear that it wouldn't be acceptable at all. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
cpufreq_add_dev() is an unnecessary wrapper over __cpufreq_add_dev(). Merge them. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Viresh Kumar authored
This clearly states what the code inside these routines is doing and how these must be used. Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org> Acked-by: Saravana Kannan <skannan@codeaurora.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 05 May, 2015 1 commit
-
-
Wang Long authored
The file 'Documentation/cpu-freq/user-guide.txt' has duplicate description of sysfs interface 'scaling_driver'. [first] scaling_driver : this file shows what cpufreq driver is used to set the frequency on this CPU [second] scaling_driver : Hardware driver for cpufreq. Although this does not affect anything, I think we should only have one. so delete the second one because the first one is described in more detail. Signed-off-by: Wang Long <long.wanglong@huawei.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
- 04 May, 2015 8 commits
-
-
Doug Smythies authored
The intel_pstate driver is difficult to debug and investigate without tsc. Also, it is likely use of tsc, and some version of C0 percentage, will be re-introdcued in futute. There have also been occasions where it is desirebale to know, and confirm, the previous target pstate. This patch brings back tsc, adds previous target pstate, and adds both to the trace data collection. Signed-off-by: Doug Smythies <dsmythies@telus.net> Acked-by: Kristen Carlson Accardi <kristen@linux.intel.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Sudeep Holla authored
The "cpu-cluster.<n>" used to get the cluster clock is not used by any platform. Moreover __of_clk_get_by_name used in clk_get return error if the "clock-names" in the DT doesn't match this string. When using DT, it's not compulsory to specify the clock name unless there are multiple clock input entries in the consumer. This patch removes the unused clock string from the driver. Acked-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Sudeep Holla authored
The actual frequency is set through "clk_change_rate" which is void function. If the underlying hardware fails and returns error, the error is lost in the clk layer. In order to track such failures, we need to read back the frequency(just the cached value as clk_recalc called after clk->ops->set_rate gets the frequency) This patch adds check to see if the frequency is set correctly or if they were any hardware failures and sends the appropriate errors to the cpufreq core. Reviewed-by: Michael Turquette <mike.turquette@linaro.org> Signed-off-by: Sudeep Holla <sudeep.holla@arm.com> Acked-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Fabian Frederick authored
pxa255_run_freqs and pxa255_turbo_freqs are only read. This patch updates arrays declaration, find_freq_tables() and its callsites. Suggested-by: Joe Perches <joe@perches.com> Signed-off-by: Fabian Frederick <fabf@skynet.be> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Fabian Frederick authored
typedef is not really useful here. Replace it by structure to improve readability. typedef should only be used in some cases. (See Documentation/CodingStyle Chapter 5 for details). Signed-off-by: Fabian Frederick <fabf@skynet.be> Acked-by: Viresh Kumar <viresh.kumar@linaro.org> Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
-
Linus Torvalds authored
-
git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4Linus Torvalds authored
Pull ext4 fixes from Ted Ts'o: "Some miscellaneous bug fixes and some final on-disk and ABI changes for ext4 encryption which provide better security and performance" * tag 'for_linus_stable' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: ext4: fix growing of tiny filesystems ext4: move check under lock scope to close a race. ext4: fix data corruption caused by unwritten and delayed extents ext4 crypto: remove duplicated encryption mode definitions ext4 crypto: do not select from EXT4_FS_ENCRYPTION ext4 crypto: add padding to filenames before encrypting ext4 crypto: simplify and speed up filename encryption
-
git://people.freedesktop.org/~airlied/linuxLinus Torvalds authored
Pull drm fixes from Dave Airlie: "One intel fix, one rockchip fix, and a bunch of radeon fixes for some regressions from audio rework and vm stability" * 'drm-fixes' of git://people.freedesktop.org/~airlied/linux: drm/i915/chv: Implement WaDisableShadowRegForCpd drm/radeon: fix userptr return value checking (v2) drm/radeon: check new address before removing old one drm/radeon: reset BOs address after clearing it. drm/radeon: fix lockup when BOs aren't part of the VM on release drm/radeon: add SI DPM quirk for Sapphire R9 270 Dual-X 2G GDDR5 drm/radeon: adjust pll when audio is not enabled drm/radeon: only enable audio streams if the monitor supports it drm/radeon: only mark audio as connected if the monitor supports it (v3) drm/radeon/audio: don't enable packets until the end drm/radeon: drop dce6_dp_enable drm/radeon: fix ordering of AVI packet setup drm/radeon: Use drm_calloc_ab for CS relocs drm/rockchip: fix error check when getting irq MAINTAINERS: add entry for Rockchip drm drivers
-
- 03 May, 2015 8 commits
-
-
git://anongit.freedesktop.org/drm-intelDave Airlie authored
Just a single intel fix * tag 'drm-intel-fixes-2015-04-30' of git://anongit.freedesktop.org/drm-intel: drm/i915/chv: Implement WaDisableShadowRegForCpd
-
https://github.com/markyzq/kernel-drm-rockchipDave Airlie authored
one fix and maintainers update * 'drm-next0420' of https://github.com/markyzq/kernel-drm-rockchip: drm/rockchip: fix error check when getting irq MAINTAINERS: add entry for Rockchip drm drivers
-
git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsiLinus Torvalds authored
Pull SCSI fixes from James Bottomley: "This is three logical fixes (as 5 patches). The 3ware class of drivers were causing an oops with multiqueue by tearing down the command mappings after completing the command (where the variables in the command used to tear down the mapping were no-longer valid). There's also a fix for the qnap iscsi target which was choking on us sending it commands that were too long and a fix for the reworked aha1542 allocating GFP_KERNEL under a lock" * tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi: 3w-9xxx: fix command completion race 3w-xxxx: fix command completion race 3w-sas: fix command completion race aha1542: Allocate memory before taking a lock SCSI: add 1024 max sectors black list flag
-
git://git.infradead.org/users/vkoul/slave-dmaLinus Torvalds authored
Pull slave dmaengine fixes from Vinod Koul: "Here are the fixes in dmaengine subsystem for rc2: - privatecnt fix for slave dma request API by Christopher - warn fix for PM ifdef in usb-dmac by Geert - fix hardware dependency for xgene by Jean" * 'next' of git://git.infradead.org/users/vkoul/slave-dma: dmaengine: increment privatecnt when using dma_get_any_slave_channel dmaengine: xgene: Set hardware dependency dmaengine: usb-dmac: Protect PM-only functions to kill warning
-
git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linuxLinus Torvalds authored
Pull powerpc fixes from Michael Ellerman: - build fix for SMP=n in book3s_xics.c - fix for Daniel's pci_controller_ops on powernv. - revert the TM syscall abort patch for now. - CPU affinity fix from Nathan. - two EEH fixes from Gavin. - fix for CR corruption from Sam. - selftest build fix. * tag 'powerpc-4.1-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mpe/linux: powerpc/powernv: Restore non-volatile CRs after nap powerpc/eeh: Delay probing EEH device during hotplug powerpc/eeh: Fix race condition in pcibios_set_pcie_reset_state() powerpc/pseries: Correct cpu affinity for dlpar added cpus selftests/powerpc: Fix the pmu install rule Revert "powerpc/tm: Abort syscalls in active transactions" powerpc/powernv: Fix early pci_controller_ops loading. powerpc/kvm: Fix SMP=n build error in book3s_xics.c
-
Jan Kara authored
The estimate of necessary transaction credits in ext4_flex_group_add() is too pessimistic. It reserves credit for sb, resize inode, and resize inode dindirect block for each group added in a flex group although they are always the same block and thus it is enough to account them only once. Also the number of modified GDT block is overestimated since we fit EXT4_DESC_PER_BLOCK(sb) descriptors in one block. Make the estimation more precise. That reduces number of requested credits enough that we can grow 20 MB filesystem (which has 1 MB journal, 79 reserved GDT blocks, and flex group size 16 by default). Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Reviewed-by: Eric Sandeen <sandeen@redhat.com>
-
Davide Italiano authored
fallocate() checks that the file is extent-based and returns EOPNOTSUPP in case is not. Other tasks can convert from and to indirect and extent so it's safe to check only after grabbing the inode mutex. Signed-off-by: Davide Italiano <dccitaliano@gmail.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Cc: stable@vger.kernel.org
-
Lukas Czerner authored
Currently it is possible to lose whole file system block worth of data when we hit the specific interaction with unwritten and delayed extents in status extent tree. The problem is that when we insert delayed extent into extent status tree the only way to get rid of it is when we write out delayed buffer. However there is a limitation in the extent status tree implementation so that when inserting unwritten extent should there be even a single delayed block the whole unwritten extent would be marked as delayed. At this point, there is no way to get rid of the delayed extents, because there are no delayed buffers to write out. So when a we write into said unwritten extent we will convert it to written, but it still remains delayed. When we try to write into that block later ext4_da_map_blocks() will set the buffer new and delayed and map it to invalid block which causes the rest of the block to be zeroed loosing already written data. For now we can fix this by simply not allowing to set delayed status on written extent in the extent status tree. Also add WARN_ON() to make sure that we notice if this happens in the future. This problem can be easily reproduced by running the following xfs_io. xfs_io -f -c "pwrite -S 0xaa 4096 2048" \ -c "falloc 0 131072" \ -c "pwrite -S 0xbb 65536 2048" \ -c "fsync" /mnt/test/fff echo 3 > /proc/sys/vm/drop_caches xfs_io -c "pwrite -S 0xdd 67584 2048" /mnt/test/fff This can be theoretically also reproduced by at random by running fsx, but it's not very reliable, though on machines with bigger page size (like ppc) this can be seen more often (especially xfstest generic/127) Signed-off-by: Lukas Czerner <lczerner@redhat.com> Signed-off-by: Theodore Ts'o <tytso@mit.edu> Cc: stable@vger.kernel.org
-