- 01 Jun, 2020 5 commits
-
-
WeiXiong Liao authored
This introduces mtdpstore, which is similar to mtdoops but more powerful. It uses pstore/blk, and aims to store panic and oops logs to a flash partition, where pstore can later read back and present as files in the mounted pstore filesystem. To make mtdpstore work, the "blkdev" of pstore/blk should be set as MTD device name or MTD device number. For more details, see Documentation/admin-guide/pstore-blk.rst This solves a number of issues: - Work duplication: both of pstore and mtdoops do the same job storing panic/oops log. They have very similar logic, registering to kmsg dumper and storing logs to several chunks one by one. - Layer violations: drivers should provides methods instead of polices. MTD should provide read/write/erase operations, and allow a higher level drivers to provide the chunk management, kmsg dump configuration, etc. - Missing features: pstore provides many additional features, including presenting the logs as files, logging dump time and count, and supporting other frontends like pmsg, console, etc. Signed-off-by:
WeiXiong Liao <liaoweixiong@allwinnertech.com> Link: https://lore.kernel.org/lkml/20200511233229.27745-11-keescook@chromium.org/ Link: https://lore.kernel.org/r/1589266715-4168-1-git-send-email-liaoweixiong@allwinnertech.comSigned-off-by:
Kees Cook <keescook@chromium.org>
-
Kees Cook authored
In order to use arbitrary block devices as a pstore backend, provide a new module param named "best_effort", which will allow using any block device, even if it has not provided a panic_write callback. Link: https://lore.kernel.org/lkml/20200511233229.27745-12-keescook@chromium.org/Signed-off-by:
-
WeiXiong Liao authored
Add support for non-block devices (e.g. MTD). A non-block driver calls pstore_blk_register_device() to register iself. In addition, pstore/zone is updated to handle non-block devices, where an erase must be done before a write. Without this, there is no way to remove records stored to an MTD. Signed-off-by:
-
WeiXiong Liao authored
In order to configure itself, the MTD backend needs to be able to query the current pstore configuration. Introduce pstore_blk_get_config() for this purpose. Signed-off-by:
-
WeiXiong Liao authored
One requirement to support MTD devices in pstore/zone is having a way to declare certain regions as broken. Add this support to pstore/zone. The MTD driver should return -ENOMSG when encountering a bad region, which tells pstore/zone to skip and try the next one. Signed-off-by:
Colin Ian King <colin.king@canonical.com> Signed-off-by:
-
- 30 May, 2020 23 commits
-
-
WeiXiong Liao authored
Add details on using pstore/blk, the new backend of pstore to record dumps to block devices, in Documentation/admin-guide/pstore-blk.rst Signed-off-by:
-
WeiXiong Liao authored
Support backend for ftrace. To enable ftrace backend, just make ftrace_size be greater than 0 and a multiple of 4096. Signed-off-by:
-
WeiXiong Liao authored
Support backend for console. To enable console backend, just make console_size be greater than 0 and a multiple of 4096. Signed-off-by:
-
WeiXiong Liao authored
Add pmsg support to pstore/blk (through pstore/zone). To enable, pmsg_size must be greater than 0 and a multiple of 4096. Signed-off-by:
-
WeiXiong Liao authored
pstore/blk is similar to pstore/ram, but uses a block device as the storage rather than persistent ram. The pstore/blk backend solves two common use-cases that used to preclude using pstore/ram: - not all devices have a battery that could be used to persist regular RAM across power failures. - most embedded intelligent equipment have no persistent ram, which increases costs, instead preferring cheaper solutions, like block devices. pstore/blk provides separate configurations for the end user and for the block drivers. User configuration determines how pstore/blk operates, such as record sizes, max kmsg dump reasons, etc. These can be set by Kconfig and/or module parameters, but module parameter have priority over Kconfig. Driver configuration covers all the details about the target block device, such as total size of the device and how to perform read/write operations. These are provided by block drivers, calling pstore_register_blkdev(), including an optional panic_write callback used to bypass regular IO APIs in an effort to avoid potentially destabilized kernel code during a panic. Signed-off-by:
-
WeiXiong Liao authored
Implement a common set of APIs needed to support pstore storage zones, based on how ramoops is designed. This will be used by pstore/blk with the intention of migrating pstore/ram in the future. Signed-off-by:
-
Pavel Tatashin authored
Currently, it is only possible to get kmsg dumps for panic and oops, or just panic, via "no-dump-oops". With "max-reason" it is possible to dump messages for other kmsg_dump events, for example emerg and shutdown. Signed-off-by:
Pavel Tatashin <pasha.tatashin@soleen.com> Link: https://lore.kernel.org/lkml/20200515184434.8470-7-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
Now that pstore_register() can correctly pass max_reason to the kmesg dump facility, introduce a new "max_reason" module parameter and "max-reason" Device Tree field. The "dump_oops" module parameter and "dump-oops" Device Tree field are now considered deprecated, but are now automatically converted to their corresponding max_reason values when present, though the new max_reason setting has precedence. For struct ramoops_platform_data, the "dump_oops" member is entirely replaced by a new "max_reason" member, with the only existing user updated in place. Additionally remove the "reason" filter logic from ramoops_pstore_write(), as that is not specifically needed anymore, though technically this is a change in behavior for any ramoops users also setting the printk.always_kmsg_dump boot param, which will cause ramoops to behave as if max_reason was set to KMSG_DUMP_MAX. Co-developed-by:
-
Pavel Tatashin authored
Add a new member to struct pstore_info for passing information about kmesg dump maximum reason. This allows a finer control of what kmesg dumps are sent to pstore storage backends. Those backends that do not explicitly set this field (keeping it equal to 0), get the default behavior: store only Oopses and Panics, or everything if the printk.always_kmsg_dump boot param is set. Signed-off-by:
-
Kees Cook authored
The pstore subsystem already had a private version of this function. With the coming addition of the pstore/zone driver, this needs to be shared. As it really should live with printk, move it there instead. Link: https://lore.kernel.org/lkml/20200515184434.8470-4-keescook@chromium.org/Acked-by:
Petr Mladek <pmladek@suse.com> Acked-by:
Sergey Senozhatsky <sergey.senozhatsky@gmail.com> Reviewed-by:
-
Pavel Tatashin authored
kmsg_dump() allows to dump kmesg buffer for various system events: oops, panic, reboot, etc. It provides an interface to register a callback call for clients, and in that callback interface there is a field "max_reason", but it was getting ignored when set to any "reason" higher than KMSG_DUMP_OOPS unless "always_kmsg_dump" was passed as kernel parameter. Allow clients to actually control their "max_reason", and keep the current behavior when "max_reason" is not set. Signed-off-by:
-
Kees Cook authored
To turn the KMSG_DUMP_* reasons into a more ordered list, collapse the redundant KMSG_DUMP_(RESTART|HALT|POWEROFF) reasons into KMSG_DUMP_SHUTDOWN. The current users already don't meaningfully distinguish between them, so there's no need to, as discussed here: https://lore.kernel.org/lkml/CA+CK2bAPv5u1ih5y9t5FUnTyximtFCtDYXJCpuyjOyHNOkRdqw@mail.gmail.com/ Link: https://lore.kernel.org/lkml/20200515184434.8470-2-keescook@chromium.org/Reviewed-by:
-
Kees Cook authored
Move the ftrace log merging logic out of pstore/ram into pstore/ftrace so other backends can use it, like pstore/zone. Link: https://lore.kernel.org/lkml/20200510202436.63222-7-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
This changes the ftrace record merging code to be agnostic of pstore/ram, as the first step to making it available as a generic routine for other backends to use, such as pstore/zone. Link: https://lore.kernel.org/lkml/20200510202436.63222-6-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
Refactor device tree size parsing routines to be able to pass a non-zero default value for providing a configurable default for the coming "max_reason" field. Also rename the helpers, since we're not always parsing a size -- we're parsing a u32 and making sure it's not greater than INT_MAX. Link: https://lore.kernel.org/lkml/20200506211523.15077-4-keescook@chromium.org/ Link: https://lore.kernel.org/lkml/20200521205223.175957-1-tyhicks@linux.microsoft.comSigned-off-by:
-
Kees Cook authored
A couple module parameters had 0600 permissions, but changing them would have no impact on ramoops, so switch these to 0400 to reflect reality. Link: https://lore.kernel.org/lkml/20200506211523.15077-7-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
It is easier to see how module params are used if they're near the variables they use. Link: https://lore.kernel.org/lkml/20200510202436.63222-4-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
If the pstore backend changes, there's no indication in the logs what the console is (it always says "pstore"). Instead, pass through the active backend's name. (Also adjust the selftest to match.) Link: https://lore.kernel.org/lkml/20200510202436.63222-5-keescook@chromium.org/ Link: https://lore.kernel.org/lkml/20200526135429.GQ12456@shao2-debianSigned-off-by:
-
Kees Cook authored
In order to more cleanly pass around backend names, make the "name" member const. This means the module param needs to be dynamic (technically, it was before, so this actually cleans up a minor memory leak if a backend was specified and then gets unloaded.) Link: https://lore.kernel.org/lkml/20200510202436.63222-3-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
The CON_ENABLED flag gets cleared during unregister_console(), so make sure we already reset the console flags before calling register_console(), otherwise unloading and reloading a pstore backend will not restart console logging. Signed-off-by: -
Kees Cook authored
If a backend was unloaded without having first removed all its associated records in pstorefs, subsequent removals would crash while attempting to call into the now missing backend. Add automatic removal from the tree in pstore_unregister(), so that no references to the backend remain. Reported-by:
Luis Henriques <lhenriques@suse.com> Link: https://lore.kernel.org/lkml/87o8yrmv69.fsf@suse.com Link: https://lore.kernel.org/lkml/20200506152114.50375-11-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
The pstore.update_ms value was being disabled during pstore_unregister(), which would cause any prior value to go unnoticed on the next pstore_register(). Instead, just let del_timer() stop the timer, which was always sufficient. This additionally refactors the timer reset code and allows the timer to be enabled if the module parameter is changed away from the default. Link: https://lore.kernel.org/lkml/20200506152114.50375-10-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
Nothing was protecting changes to the pstorefs superblock. Add locking and refactor away is_pstore_mounted(), instead using a helper to add a way to safely lock the pstorefs root inode during filesystem changes. Link: https://lore.kernel.org/lkml/20200506152114.50375-9-keescook@chromium.org/Signed-off-by:
-
- 12 May, 2020 7 commits
-
-
Kees Cook authored
The "unlink" handling should perform list removal (which can also make sure records don't get double-erased), and the "evict" handling should be responsible only for memory freeing. Link: https://lore.kernel.org/lkml/20200506152114.50375-8-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
The pstore backend lock wasn't being used during pstore_unregister(). Add sanity check and locking. Link: https://lore.kernel.org/lkml/20200506152114.50375-7-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
The pstorefs internal list lock doesn't need to be a spinlock and will create problems when trying to access the list in the subsequent patch that will walk the pstorefs records during pstore_unregister(). Change this to a mutex to avoid may_sleep() warnings when unregistering devices. Link: https://lore.kernel.org/lkml/20200506152114.50375-6-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
The name "allpstore" doesn't carry much meaning, so rename it to what it actually is: the list of all records present in the filesystem. The lock is also renamed accordingly. Link: https://lore.kernel.org/lkml/20200506152114.50375-5-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
Currently pstore can only have a single backend attached at a time, and it tracks the active backend via "psinfo", under a lock. The locking for this does not need to be a spinlock, and in order to avoid may_sleep() issues during future changes to pstore_unregister(), switch to a mutex instead. Link: https://lore.kernel.org/lkml/20200506152114.50375-4-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
The name "pstore_lock" sounds very global, but it is only supposed to be used for managing changes to "psinfo", so rename it accordingly. Link: https://lore.kernel.org/lkml/20200506152114.50375-3-keescook@chromium.org/Signed-off-by:
-
Kees Cook authored
There is no reason to be doing a module get/put in pstore_register(), since the module calling pstore_register() cannot be unloaded since it hasn't finished its initialization. Remove it so there is no confusion about how registration ordering works. Link: https://lore.kernel.org/lkml/20200506152114.50375-2-keescook@chromium.org/Signed-off-by:
-
- 12 Apr, 2020 5 commits
-
-
Linus Torvalds authored
-
Linus Torvalds authored
This sorts the actual field names too, potentially causing even more chaos and confusion at merge time if you have edited the MAINTAINERS file. But the end result is a more consistent layout, and hopefully it's a one-time pain minimized by doing this just before the -rc1 release. This was entirely scripted: ./scripts/parse-maintainers.pl --input=MAINTAINERS --output=MAINTAINERS --order Requested-by:
Joe Perches <joe@perches.com> Signed-off-by:
Linus Torvalds <torvalds@linux-foundation.org>
-
Linus Torvalds authored
They are all supposed to be sorted, but people who add new entries don't always know the alphabet. Plus sometimes the entry names get edited, and people don't then re-order the entry. Let's see how painful this will be for merging purposes (the MAINTAINERS file is often edited in various different trees), but Joe claims there's relatively few patches in -next that touch this, and doing it just before -rc1 is likely the best time. Fingers crossed. This was scripted with /scripts/parse-maintainers.pl --input=MAINTAINERS --output=MAINTAINERS but then I also ended up manually upper-casing a few entry names that stood out when looking at the end result. Requested-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tipLinus Torvalds authored
Pull x86 fixes from Thomas Gleixner: "A set of three patches to fix the fallout of the newly added split lock detection feature. It addressed the case where a KVM guest triggers a split lock #AC and KVM reinjects it into the guest which is not prepared to handle it. Add proper sanity checks which prevent the unconditional injection into the guest and handles the #AC on the host side in the same way as user space detections are handled. Depending on the detection mode it either warns and disables detection for the task or kills the task if the mode is set to fatal" * tag 'x86-urgent-2020-04-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: KVM: VMX: Extend VMXs #AC interceptor to handle split lock #AC in guest KVM: x86: Emulate split-lock access as a write in emulator x86/split_lock: Provide handle_guest_split_lock()
-
git://git.kernel.org/pub/scm/linux/kernel/git/tip/tipLinus Torvalds authored
Pull time(keeping) updates from Thomas Gleixner: - Fix the time_for_children symlink in /proc/$PID/ so it properly reflects that it part of the 'time' namespace - Add the missing userns limit for the allowed number of time namespaces, which was half defined but the actual array member was not added. This went unnoticed as the array has an exessive empty member at the end but introduced a user visible regression as the output was corrupted. - Prevent further silent ucount corruption by adding a BUILD_BUG_ON() to catch half updated data. * tag 'timers-urgent-2020-04-12' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: ucount: Make sure ucounts in /proc/sys/user don't regress again time/namespace: Add max_time_namespaces ucount time/namespace: Fix time_for_children symlink
-
