linux-block.git
5 months agoloop: move updating lo_flags out of loop_set_status_from_info
Christoph Hellwig [Fri, 10 Jan 2025 07:37:31 +0000 (08:37 +0100)]
loop: move updating lo_flags out of loop_set_status_from_info

While loop_configure simplify assigns the flags passed in by userspace,
loop_set_status only looks at the two changeable flags, and currently
has to do a complicate dance to implement that.

Move assign lo->lo_flags out of loop_set_status_from_info into the
callers and thus drastically simplify the lo_flags handling in
loop_set_status.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250110073750.1582447-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoloop: fix queue freeze vs limits lock order
Christoph Hellwig [Fri, 10 Jan 2025 05:47:19 +0000 (06:47 +0100)]
loop: fix queue freeze vs limits lock order

Match the locking order used by the core block code by only freezing
the queue after taking the limits lock using the
queue_limits_commit_update_frozen helper and document the callers that
do not freeze the queue at all.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-12-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoloop: refactor queue limits updates
Christoph Hellwig [Fri, 10 Jan 2025 05:47:18 +0000 (06:47 +0100)]
loop: refactor queue limits updates

Replace loop_reconfigure_limits with a slightly less encompassing
loop_update_limits that expects the caller to acquire and commit the
queue limits to prepare for sorting out the freeze vs limits lock
ordering.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-11-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agousb-storage: fix queue freeze vs limits lock order
Christoph Hellwig [Fri, 10 Jan 2025 05:47:17 +0000 (06:47 +0100)]
usb-storage: fix queue freeze vs limits lock order

Match the locking order used by the core block code by only freezing
the queue after taking the limits lock using the
queue_limits_commit_update_frozen helper.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-10-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agonbd: fix queue freeze vs limits lock order
Christoph Hellwig [Fri, 10 Jan 2025 05:47:16 +0000 (06:47 +0100)]
nbd: fix queue freeze vs limits lock order

Match the locking order used by the core block code by only freezing
the queue after taking the limits lock using the
queue_limits_commit_update_frozen helper.

This also allows removes the need for the separate __nbd_set_size helper,
so remove it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-9-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agonvme: fix queue freeze vs limits lock order
Christoph Hellwig [Fri, 10 Jan 2025 05:47:15 +0000 (06:47 +0100)]
nvme: fix queue freeze vs limits lock order

Match the locking order used by the core block code by only freezing
the queue after taking the limits lock.

Unlike most queue updates this does not use the
queue_limits_commit_update_frozen helper as the nvme driver want the
queue frozen for more than just the limits update.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-8-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: fix queue freeze vs limits lock order in sysfs store methods
Christoph Hellwig [Fri, 10 Jan 2025 05:47:14 +0000 (06:47 +0100)]
block: fix queue freeze vs limits lock order in sysfs store methods

queue_attr_store() always freezes a device queue before calling the
attribute store operation. For attributes that control queue limits, the
store operation will also lock the queue limits with a call to
queue_limits_start_update(). However, some drivers (e.g. SCSI sd) may
need to issue commands to a device to obtain limit values from the
hardware with the queue limits locked. This creates a potential ABBA
deadlock situation if a user attempts to modify a limit (thus freezing
the device queue) while the device driver starts a revalidation of the
device queue limits.

Avoid such deadlock by not freezing the queue before calling the
->store_limit() method in struct queue_sysfs_entry and instead use the
queue_limits_commit_update_frozen helper to freeze the queue after taking
the limits lock.

This also removes taking the sysfs lock for the store_limit method as
it doesn't protect anything here, but creates even more nesting.
Hopefully it will go away from the actual sysfs methods entirely soon.

(commit log adapted from a similar patch from  Damien Le Moal)

Fixes: ff956a3be95b ("block: use queue_limits_commit_update in queue_discard_max_store")
Fixes: 0327ca9d53bf ("block: use queue_limits_commit_update in queue_max_sectors_store")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-7-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: add a store_limit operations for sysfs entries
Christoph Hellwig [Fri, 10 Jan 2025 05:47:13 +0000 (06:47 +0100)]
block: add a store_limit operations for sysfs entries

De-duplicate the code for updating queue limits by adding a store_limit
method that allows having common code handle the actual queue limits
update.

Note that this is a pure refactoring patch and does not address the
existing freeze vs limits lock order problem in the refactored code,
which will be addressed next.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-6-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: don't update BLK_FEAT_POLL in __blk_mq_update_nr_hw_queues
Christoph Hellwig [Fri, 10 Jan 2025 05:47:12 +0000 (06:47 +0100)]
block: don't update BLK_FEAT_POLL in __blk_mq_update_nr_hw_queues

When __blk_mq_update_nr_hw_queues changes the number of tag sets, it
might have to disable poll queues.  Currently it does so by adjusting
the BLK_FEAT_POLL, which is a bit against the intent of features that
describe hardware / driver capabilities, but more importantly causes
nasty lock order problems with the broadly held freeze when updating the
number of hardware queues and the limits lock.  Fix this by leaving
BLK_FEAT_POLL alone, and instead check for the number of poll queues in
the bio submission and poll handlers.  While this adds extra work to the
fast path, the variables are in cache lines used by these operations
anyway, so it should be cheap enough.

Fixes: 8023e144f9d6 ("block: move the poll flag to queue_limits")
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: check BLK_FEAT_POLL under q_usage_count
Christoph Hellwig [Fri, 10 Jan 2025 05:47:11 +0000 (06:47 +0100)]
block: check BLK_FEAT_POLL under q_usage_count

Otherwise feature reconfiguration can race with I/O submission.

Also drop the bio_clear_polled in the error path, as the flag does not
matter for instant error completions, it is a left over from when we
allowed polled I/O to proceed unpolled in this case.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: add a queue_limits_commit_update_frozen helper
Christoph Hellwig [Fri, 10 Jan 2025 05:47:10 +0000 (06:47 +0100)]
block: add a queue_limits_commit_update_frozen helper

Add a helper that freezes the queue, updates the queue limits and
unfreezes the queue and convert all open coded versions of that to the
new helper.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: fix docs for freezing of queue limits updates
Christoph Hellwig [Fri, 10 Jan 2025 05:47:09 +0000 (06:47 +0100)]
block: fix docs for freezing of queue limits updates

queue_limits_commit_update is the function that needs to operate on a
frozen queue, not queue_limits_start_update.  Update the kerneldoc
comments to reflect that.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Nilay Shroff <nilay@linux.ibm.com>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Link: https://lore.kernel.org/r/20250110054726.1499538-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agonbd: don't allow reconnect after disconnect
Yu Kuai [Fri, 3 Jan 2025 09:28:59 +0000 (17:28 +0800)]
nbd: don't allow reconnect after disconnect

Following process can cause nbd_config UAF:

1) grab nbd_config temporarily;

2) nbd_genl_disconnect() flush all recv_work() and release the
initial reference:

  nbd_genl_disconnect
   nbd_disconnect_and_put
    nbd_disconnect
     flush_workqueue(nbd->recv_workq)
    if (test_and_clear_bit(NBD_RT_HAS_CONFIG_REF, ...))
     nbd_config_put
     -> due to step 1), reference is still not zero

3) nbd_genl_reconfigure() queue recv_work() again;

  nbd_genl_reconfigure
   config = nbd_get_config_unlocked(nbd)
   if (!config)
   -> succeed
   if (!test_bit(NBD_RT_BOUND, ...))
   -> succeed
   nbd_reconnect_socket
    queue_work(nbd->recv_workq, &args->work)

4) step 1) release the reference;

5) Finially, recv_work() will trigger UAF:

  recv_work
   nbd_config_put(nbd)
   -> nbd_config is freed
   atomic_dec(&config->recv_threads)
   -> UAF

Fix the problem by clearing NBD_RT_BOUND in nbd_genl_disconnect(), so
that nbd_genl_reconfigure() will fail.

Fixes: b7aa3d39385d ("nbd: add a reconfigure netlink command")
Reported-by: syzbot+6b0df248918b92c33e6a@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/all/675bfb65.050a0220.1a2d0d.0006.GAE@google.com/
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250103092859.3574648-1-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: simplify tag allocation policy selection
Christoph Hellwig [Mon, 6 Jan 2025 08:35:11 +0000 (09:35 +0100)]
block: simplify tag allocation policy selection

Use a plain BLK_MQ_F_* flag to select the round robin tag selection
instead of overlaying an enum with just two possible values into the
flags space.

Doing so allows adding a BLK_MQ_F_MAX sentinel for simplified overflow
checking in the messy debugfs helpers.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Link: https://lore.kernel.org/r/20250106083531.799976-5-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: remove BLK_MQ_F_NO_SCHED
Christoph Hellwig [Mon, 6 Jan 2025 08:35:10 +0000 (09:35 +0100)]
block: remove BLK_MQ_F_NO_SCHED

The only queues that really can't support a scheduler are those that
do not have a gendisk associated with them, and thus can't be used for
non-passthrough commands.  In addition to those null_blk can optionally
set the flag, which is a bad odd.  Replace the null_blk usage with
BLK_MQ_F_NO_SCHED_BY_DEFAULT to keep the expected semantics and then
remove BLK_MQ_F_NO_SCHED as the non-disk queues never call into
elevator_init_mq or blk_register_queue which adds the sysfs attributes.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250106083531.799976-4-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: remove blk_mq_init_bitmaps
Christoph Hellwig [Mon, 6 Jan 2025 08:35:09 +0000 (09:35 +0100)]
block: remove blk_mq_init_bitmaps

The little work done in blk_mq_init_bitmaps is easier done in the only
caller.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Link: https://lore.kernel.org/r/20250106083531.799976-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: better split mq vs non-mq code in add_disk_fwnode
Christoph Hellwig [Mon, 6 Jan 2025 08:35:08 +0000 (09:35 +0100)]
block: better split mq vs non-mq code in add_disk_fwnode

Add a big conditional for blk-mq vs not mq at the beginning of
add_disk_fwnode so that elevator_init_mq is only called for blk-mq disks,
and add checks that the right methods or set or not set based on the
queue type.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250106083531.799976-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: add a dma mapping iterator
Christoph Hellwig [Mon, 6 Jan 2025 08:15:29 +0000 (09:15 +0100)]
block: add a dma mapping iterator

blk_rq_map_sg is maze of nested loops.  Untangle it by creating an
iterator that returns [paddr,len] tuples for DMA mapping, and then
implement the DMA logic on top of this.  This not only removes code
at the source level, but also generates nicer binary code:

$ size block/blk-merge.o.*
   text    data     bss     dec     hex filename
  10001     432       0   10433    28c1 block/blk-merge.o.new
  10317     468       0   10785    2a21 block/blk-merge.o.old

Last but not least it will be used as a building block for a new
DMA mapping helper that doesn't rely on struct scatterlist.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250106081609.798289-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: use page_to_phys in bvec_phys
Christoph Hellwig [Mon, 6 Jan 2025 08:14:37 +0000 (09:14 +0100)]
block: use page_to_phys in bvec_phys

Use page_to_phys instead of open coding it now that it is available in an
architecture independent way.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20250106081437.798213-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: remove blk_rq_bio_prep
Christoph Hellwig [Fri, 3 Jan 2025 07:33:58 +0000 (08:33 +0100)]
block: remove blk_rq_bio_prep

There is not real point in a helper just to assign three values to four
fields, especially when the surrounding code is working on the
neighbor fields directly.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Link: https://lore.kernel.org/r/20250103073417.459715-3-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: remove bio_add_pc_page
Christoph Hellwig [Fri, 3 Jan 2025 07:33:57 +0000 (08:33 +0100)]
block: remove bio_add_pc_page

Lift bio_split_rw_at into blk_rq_append_bio so that it validates the
hardware limits.  With this all passthrough callers can simply add
bio_add_page to build the bio and delay checking for exceeding of limits
to this point instead of doing it for each page.

While this looks like adding a new expensive loop over all bio_vecs,
blk_rq_append_bio is already doing that just to counter the number of
segments.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Link: https://lore.kernel.org/r/20250103073417.459715-2-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agops3disk: Do not use dev->bounce_size before it is set
Geert Uytterhoeven [Fri, 3 Jan 2025 08:51:25 +0000 (09:51 +0100)]
ps3disk: Do not use dev->bounce_size before it is set

dev->bounce_size is only initialized after it is used to set the queue
limits.  Fix this by using BOUNCE_SIZE instead.

Fixes: a7f18b74dbe17162 ("ps3disk: pass queue_limits to blk_mq_alloc_disk")
Reported-by: Philipp Hortmann <philipp.g.hortmann@gmail.com>
Closes: https://lore.kernel.org/39256db9-3d73-4e86-a49b-300dfd670212@gmail.com
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/06988f959ea6885b8bd7fb3b9059dd54bc6bbad7.1735894216.git.geert+renesas@glider.be
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: retry call probe after request_module in blk_request_module
Yang Erkun [Mon, 9 Dec 2024 11:04:35 +0000 (19:04 +0800)]
block: retry call probe after request_module in blk_request_module

Set kernel config:

 CONFIG_BLK_DEV_LOOP=m
 CONFIG_BLK_DEV_LOOP_MIN_COUNT=0

Do latter:

 mknod loop0 b 7 0
 exec 4<> loop0

Before commit e418de3abcda ("block: switch gendisk lookup to a simple
xarray"), lookup_gendisk will first use base_probe to load module loop,
and then the retry will call loop_probe to prepare the loop disk. Finally
open for this disk will success. However, after this commit, we lose the
retry logic, and open will fail with ENXIO. Block device autoloading is
deprecated and will be removed soon, but maybe we should keep open success
until we really remove it. So, give a retry to fix it.

Fixes: e418de3abcda ("block: switch gendisk lookup to a simple xarray")
Suggested-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Yang Erkun <yangerkun@huawei.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20241209110435.3670985-1-yangerkun@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agokyber: constify sysfs attributes
Thomas Weißschuh [Thu, 2 Jan 2025 12:01:34 +0000 (13:01 +0100)]
kyber: constify sysfs attributes

The elevator core now allows instances of 'struct elv_fs_entry' to be
moved into read-only memory. Make use of that to protect them against
accidental or malicious modifications.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20250102-sysfs-const-attr-elevator-v1-4-9837d2058c60@weissschuh.net
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock, bfq: constify sysfs attributes
Thomas Weißschuh [Thu, 2 Jan 2025 12:01:33 +0000 (13:01 +0100)]
block, bfq: constify sysfs attributes

The elevator core now allows instances of 'struct elv_fs_entry' to be
moved into read-only memory. Make use of that to protect them against
accidental or malicious modifications.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20250102-sysfs-const-attr-elevator-v1-3-9837d2058c60@weissschuh.net
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: mq-deadline: Constify sysfs attributes
Thomas Weißschuh [Thu, 2 Jan 2025 12:01:32 +0000 (13:01 +0100)]
block: mq-deadline: Constify sysfs attributes

The elevator core now allows instances of 'struct elv_fs_entry' to be
moved into read-only memory. Make use of that to protect them against
accidental or malicious modifications.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20250102-sysfs-const-attr-elevator-v1-2-9837d2058c60@weissschuh.net
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoelevator: Enable const sysfs attributes
Thomas Weißschuh [Thu, 2 Jan 2025 12:01:31 +0000 (13:01 +0100)]
elevator: Enable const sysfs attributes

The elevator core does not need to modify the sysfs attributes added by
the elevators. Reflect this in the types, so the attributes can be moved
into read-only memory.

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20250102-sysfs-const-attr-elevator-v1-1-9837d2058c60@weissschuh.net
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblk-zoned: Split queue_zone_wplugs_show()
Bart Van Assche [Tue, 17 Dec 2024 21:03:10 +0000 (13:03 -0800)]
blk-zoned: Split queue_zone_wplugs_show()

Reduce the indentation level of the code in queue_zone_wplugs_show() by
moving the body of the loop in that function into a new function.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20241217210310.645966-5-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblk-zoned: Improve the queue reference count strategy documentation
Bart Van Assche [Tue, 17 Dec 2024 21:03:09 +0000 (13:03 -0800)]
blk-zoned: Improve the queue reference count strategy documentation

For the blk_queue_exit() calls, document where the corresponding code can
be found that increases q->q_usage_counter.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20241217210310.645966-4-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblk-zoned: Document locking assumptions
Bart Van Assche [Tue, 17 Dec 2024 21:03:08 +0000 (13:03 -0800)]
blk-zoned: Document locking assumptions

Document which functions expect that their callers must hold a lock.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20241217210310.645966-3-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblk-zoned: Minimize #include directives
Bart Van Assche [Tue, 17 Dec 2024 21:03:07 +0000 (13:03 -0800)]
blk-zoned: Minimize #include directives

Only include those header files that are necessary.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
Link: https://lore.kernel.org/r/20241217210310.645966-2-bvanassche@acm.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agorust: block: fix use of BLK_MQ_F_SHOULD_MERGE
Andreas Hindborg [Fri, 20 Dec 2024 09:37:57 +0000 (10:37 +0100)]
rust: block: fix use of BLK_MQ_F_SHOULD_MERGE

BLK_MQ_F_SHOULD_MERGE has was removed [1] and is now in effect by default.
So remove the flag from tag sets of Rust block device drivers.

Link: https://lore.kernel.org/r/20241219060214.1928848-1-hch@lst.de
Fixes: 9377b95cda73 ("block: remove BLK_MQ_F_SHOULD_MERGE")
Signed-off-by: Andreas Hindborg <a.hindborg@kernel.org>
Link: https://lore.kernel.org/r/20241220-merge-flag-fix-v1-1-41b7778dac06@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: remove BLK_MQ_F_SHOULD_MERGE
Christoph Hellwig [Thu, 19 Dec 2024 06:01:59 +0000 (07:01 +0100)]
block: remove BLK_MQ_F_SHOULD_MERGE

BLK_MQ_F_SHOULD_MERGE is set for all tag_sets except those that purely
process passthrough commands (bsg-lib, ufs tmf, various nvme admin
queues) and thus don't even check the flag.  Remove it to simplify the
driver interface.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Link: https://lore.kernel.org/r/20241219060214.1928848-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblk-mq: remove unused queue mapping helpers
Daniel Wagner [Mon, 2 Dec 2024 14:00:16 +0000 (15:00 +0100)]
blk-mq: remove unused queue mapping helpers

There are no users left of the pci and virtio queue mapping helpers.
Thus remove them.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-8-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agovirtio: blk/scsi: replace blk_mq_virtio_map_queues with blk_mq_map_hw_queues
Daniel Wagner [Mon, 2 Dec 2024 14:00:15 +0000 (15:00 +0100)]
virtio: blk/scsi: replace blk_mq_virtio_map_queues with blk_mq_map_hw_queues

Replace all users of blk_mq_virtio_map_queues with the more generic
blk_mq_map_hw_queues. This in preparation to retire
blk_mq_virtio_map_queues.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-7-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agonvme: replace blk_mq_pci_map_queues with blk_mq_map_hw_queues
Daniel Wagner [Mon, 2 Dec 2024 14:00:14 +0000 (15:00 +0100)]
nvme: replace blk_mq_pci_map_queues with blk_mq_map_hw_queues

Replace all users of blk_mq_pci_map_queues with the more generic
blk_mq_map_hw_queues. This in preparation to retire
blk_mq_pci_map_queues.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-6-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoscsi: replace blk_mq_pci_map_queues with blk_mq_map_hw_queues
Daniel Wagner [Mon, 2 Dec 2024 14:00:13 +0000 (15:00 +0100)]
scsi: replace blk_mq_pci_map_queues with blk_mq_map_hw_queues

Replace all users of blk_mq_pci_map_queues with the more generic
blk_mq_map_hw_queues. This in preparation to retire
blk_mq_pci_map_queues.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-5-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblk-mq: introduce blk_mq_map_hw_queues
Daniel Wagner [Mon, 2 Dec 2024 14:00:12 +0000 (15:00 +0100)]
blk-mq: introduce blk_mq_map_hw_queues

blk_mq_pci_map_queues and blk_mq_virtio_map_queues will create a CPU to
hardware queue mapping based on affinity information. These two function
share common code and only differ on how the affinity information is
retrieved. Also, those functions are located in the block subsystem
where it doesn't really fit in. They are virtio and pci subsystem
specific.

Thus introduce provide a generic mapping function which uses the
irq_get_affinity callback from bus_type.

Originally idea from Ming Lei <ming.lei@redhat.com>

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-4-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agovirtio: hookup irq_get_affinity callback
Daniel Wagner [Mon, 2 Dec 2024 14:00:11 +0000 (15:00 +0100)]
virtio: hookup irq_get_affinity callback

struct bus_type has a new callback for retrieving the IRQ affinity for a
device. Hook this callback up for virtio based devices.

Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-3-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoPCI: hookup irq_get_affinity callback
Daniel Wagner [Mon, 2 Dec 2024 14:00:10 +0000 (15:00 +0100)]
PCI: hookup irq_get_affinity callback

struct bus_type has a new callback for retrieving the IRQ affinity for a
device. Hook this callback up for PCI based devices.

Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-2-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agodriver core: bus: add irq_get_affinity callback to bus_type
Daniel Wagner [Mon, 2 Dec 2024 14:00:09 +0000 (15:00 +0100)]
driver core: bus: add irq_get_affinity callback to bus_type

Introducing a callback in struct bus_type so that a subsystem
can hook up the getters directly. This approach avoids exposing
random getters in any subsystems APIs.

Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Daniel Wagner <wagi@kernel.org>
Link: https://lore.kernel.org/r/20241202-refactor-blk-affinity-helpers-v6-1-27211e9c2cd5@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agonull_blk: Remove accesses to page->index
Matthew Wilcox (Oracle) [Mon, 16 Dec 2024 16:08:47 +0000 (16:08 +0000)]
null_blk: Remove accesses to page->index

Use page->private to store the index instead of page->index.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://lore.kernel.org/r/20241216160849.31739-1-willy@infradead.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: rnull: Initialize the module in place
Benoît du Garreau [Wed, 4 Dec 2024 08:38:39 +0000 (09:38 +0100)]
block: rnull: Initialize the module in place

Using `InPlaceModule` avoids an allocation and an indirection.

Signed-off-by: Benoît du Garreau <benoit@dugarreau.fr>
Acked-by: Andreas Hindborg <a.hindborg@kernel.org>
Link: https://lore.kernel.org/r/20241204-rnull_in_place-v1-1-efe3eafac9fb@dugarreau.fr
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblktrace: remove redundant return at end of function
Colin Ian King [Wed, 4 Dec 2024 15:04:50 +0000 (15:04 +0000)]
blktrace: remove redundant return at end of function

A recent change added return 0 before an existing return statement
at the end of function blk_trace_setup. The final return is now
redundant, so remove it.

Fixes: 64d124798244 ("blktrace: move copy_[to|from]_user() out of ->debugfs_lock")
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20241204150450.399005-1-colin.i.king@gmail.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: Delete bio_set_prio()
John Garry [Mon, 2 Dec 2024 11:19:57 +0000 (11:19 +0000)]
block: Delete bio_set_prio()

Since commit 43b62ce3ff0a ("block: move bio io prio to a new field"), macro
bio_set_prio() does nothing but set bio->bi_ioprio. All other places just
set bio->bi_ioprio directly, so replace bio_set_prio() remaining
callsites with setting bio->bi_ioprio directly and delete that macro.

Signed-off-by: John Garry <john.g.garry@oracle.com>
Acked-by: Jack Wang <jinpu.wang@ionos.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20241202111957.2311683-3-john.g.garry@oracle.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: Delete bio_prio()
John Garry [Mon, 2 Dec 2024 11:19:56 +0000 (11:19 +0000)]
block: Delete bio_prio()

Since commit 43b62ce3ff0a ("block: move bio io prio to a new field"), macro
bio_prio() does nothing but return the value in bio->bi_ioprio. Most other
places just read bio->bi_ioprio directly, so replace bi_ioprio() callsites
with reading bio->bi_ioprio directly and delete that macro.

Signed-off-by: John Garry <john.g.garry@oracle.com>
Reviewed-by: Chaitanya Kulkarni <kch@nvidia.com>
Link: https://lore.kernel.org/r/20241202111957.2311683-2-john.g.garry@oracle.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblktrace: move copy_[to|from]_user() out of ->debugfs_lock
Ming Lei [Thu, 28 Nov 2024 12:50:27 +0000 (20:50 +0800)]
blktrace: move copy_[to|from]_user() out of ->debugfs_lock

Move copy_[to|from]_user() out of ->debugfs_lock and cut the dependency
between mm->mmap_lock and q->debugfs_lock, then we avoids lots of
lockdep false positive warning. Obviously ->debug_lock isn't needed
for copy_[to|from]_user().

The only behavior change is to call blk_trace_remove() in case of setup
failure handling by re-grabbing ->debugfs_lock, and this way is just
fine since we do cover concurrent setup() & remove().

Reported-by: syzbot+91585b36b538053343e4@syzkaller.appspotmail.com
Closes: https://lore.kernel.org/linux-block/67450fd4.050a0220.1286eb.0007.GAE@google.com/
Closes: https://lore.kernel.org/linux-block/6742e584.050a0220.1cc393.0038.GAE@google.com/
Closes: https://lore.kernel.org/linux-block/6742a600.050a0220.1cc393.002e.GAE@google.com/
Closes: https://lore.kernel.org/linux-block/67420102.050a0220.1cc393.0019.GAE@google.com/
Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20241128125029.4152292-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblktrace: don't centralize grabbing q->debugfs_mutex in blk_trace_ioctl
Ming Lei [Thu, 28 Nov 2024 12:50:26 +0000 (20:50 +0800)]
blktrace: don't centralize grabbing q->debugfs_mutex in blk_trace_ioctl

Call each handler directly and the handler do grab q->debugfs_mutex,
prepare for killing dependency between ->debug_mutex and ->mmap_lock.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20241128125029.4152292-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agonull_blk: Add rotational feature support
Damien Le Moal [Tue, 26 Nov 2024 00:09:56 +0000 (09:09 +0900)]
null_blk: Add rotational feature support

To facilitate testing of kernel functions related to the rotational
feature (BLK_FEAT_ROTATIONAL) of a block device (e.g. NVMe rotational
bit support), add the rotational boolean configfs attribute and module
parameter to the null_blk driver. If set, a null block device will
report being a rotational device through it queue limits features with
the BLK_FEAT_ROTATIONAL flag.

Signed-off-by: Damien Le Moal <dlemoal@kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/20241126000956.95983-1-dlemoal@kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: track queue dying state automatically for modeling queue freeze lockdep
Ming Lei [Wed, 27 Nov 2024 13:51:30 +0000 (21:51 +0800)]
block: track queue dying state automatically for modeling queue freeze lockdep

Now we only verify the outmost freeze & unfreeze in current context in case
that !q->mq_freeze_depth, so it is reliable to save queue lying state when
we want to lock the freeze queue since the state is one per-task variable
now.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20241127135133.3952153-5-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: don't verify queue freeze manually in elevator_init_mq()
Ming Lei [Wed, 27 Nov 2024 13:51:29 +0000 (21:51 +0800)]
block: don't verify queue freeze manually in elevator_init_mq()

Now blk_freeze_queue_start() can track disk state automatically, and
it isn't necessary to verify queue freeze manually in elevator_init_mq()
any more.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20241127135133.3952153-4-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: track disk DEAD state automatically for modeling queue freeze lockdep
Ming Lei [Wed, 27 Nov 2024 13:51:28 +0000 (21:51 +0800)]
block: track disk DEAD state automatically for modeling queue freeze lockdep

Now we only verify the outmost freeze & unfreeze in current context in case
that !q->mq_freeze_depth, so it is reliable to save disk DEAD state when
we want to lock the freeze queue since the state is one per-task variable
now.

Doing this way can kill lots of false positive when freeze queue is
called before adding disk[1].

[1] https://lore.kernel.org/linux-block/6741f6b2.050a0220.1cc393.0017.GAE@google.com/

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20241127135133.3952153-3-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoblock: remove unnecessary check in blk_unfreeze_check_owner()
Ming Lei [Wed, 27 Nov 2024 13:51:27 +0000 (21:51 +0800)]
block: remove unnecessary check in blk_unfreeze_check_owner()

The following check of 'q->mq_freeze_owner != current' covers the
previous one, so remove the unnecessary check.

Signed-off-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20241127135133.3952153-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 months agoLinux 6.13-rc4 v6.13-rc4
Linus Torvalds [Sun, 22 Dec 2024 21:22:21 +0000 (13:22 -0800)]
Linux 6.13-rc4

5 months agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Sun, 22 Dec 2024 20:16:41 +0000 (12:16 -0800)]
Merge tag 'for-linus' of git://git./virt/kvm/kvm

Pull KVM x86 fixes from Paolo Bonzini:

 - Disable AVIC on SNP-enabled systems that don't allow writes to the
   virtual APIC page, as such hosts will hit unexpected RMP #PFs in the
   host when running VMs of any flavor.

 - Fix a WARN in the hypercall completion path due to KVM trying to
   determine if a guest with protected register state is in 64-bit mode
   (KVM's ABI is to assume such guests only make hypercalls in 64-bit
   mode).

 - Allow the guest to write to supported bits in MSR_AMD64_DE_CFG to fix
   a regression with Windows guests, and because KVM's read-only
   behavior appears to be entirely made up.

 - Treat TDP MMU faults as spurious if the faulting access is allowed
   given the existing SPTE. This fixes a benign WARN (other than the
   WARN itself) due to unexpectedly replacing a writable SPTE with a
   read-only SPTE.

 - Emit a warning when KVM is configured with ignore_msrs=1 and also to
   hide the MSRs that the guest is looking for from the kernel logs.
   ignore_msrs can trick guests into assuming that certain processor
   features are present, and this in turn leads to bogus bug reports.

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
  KVM: x86: let it be known that ignore_msrs is a bad idea
  KVM: VMX: don't include '<linux/find.h>' directly
  KVM: x86/mmu: Treat TDP MMU faults as spurious if access is already allowed
  KVM: SVM: Allow guest writes to set MSR_AMD64_DE_CFG bits
  KVM: x86: Play nice with protected guests in complete_hypercall_exit()
  KVM: SVM: Disable AVIC on SNP-enabled system without HvInUseWrAllowed feature

5 months agoMerge tag 'kvm-x86-fixes-6.13-rcN' of https://github.com/kvm-x86/linux into HEAD
Paolo Bonzini [Sun, 22 Dec 2024 17:07:16 +0000 (12:07 -0500)]
Merge tag 'kvm-x86-fixes-6.13-rcN' of https://github.com/kvm-x86/linux into HEAD

KVM x86 fixes for 6.13:

 - Disable AVIC on SNP-enabled systems that don't allow writes to the virtual
   APIC page, as such hosts will hit unexpected RMP #PFs in the host when
   running VMs of any flavor.

 - Fix a WARN in the hypercall completion path due to KVM trying to determine
   if a guest with protected register state is in 64-bit mode (KVM's ABI is to
   assume such guests only make hypercalls in 64-bit mode).

 - Allow the guest to write to supported bits in MSR_AMD64_DE_CFG to fix a
   regression with Windows guests, and because KVM's read-only behavior appears
   to be entirely made up.

 - Treat TDP MMU faults as spurious if the faulting access is allowed given the
   existing SPTE.  This fixes a benign WARN (other than the WARN itself) due to
   unexpectedly replacing a writable SPTE with a read-only SPTE.

5 months agoKVM: x86: let it be known that ignore_msrs is a bad idea
Paolo Bonzini [Thu, 19 Dec 2024 12:43:20 +0000 (07:43 -0500)]
KVM: x86: let it be known that ignore_msrs is a bad idea

When running KVM with ignore_msrs=1 and report_ignored_msrs=0, the user has
no clue that that the guest is being lied to.  This may cause bug reports
such as https://gitlab.com/qemu-project/qemu/-/issues/2571, where enabling
a CPUID bit in QEMU caused Linux guests to try reading MSR_CU_DEF_ERR; and
being lied about the existence of MSR_CU_DEF_ERR caused the guest to assume
other things about the local APIC which were not true:

  Sep 14 12:02:53 kernel: mce: [Firmware Bug]: Your BIOS is not setting up LVT offset 0x2 for deferred error IRQs correctly.
  Sep 14 12:02:53 kernel: unchecked MSR access error: RDMSR from 0x852 at rIP: 0xffffffffb548ffa7 (native_read_msr+0x7/0x40)
  Sep 14 12:02:53 kernel: Call Trace:
  ...
  Sep 14 12:02:53 kernel:  native_apic_msr_read+0x20/0x30
  Sep 14 12:02:53 kernel:  setup_APIC_eilvt+0x47/0x110
  Sep 14 12:02:53 kernel:  mce_amd_feature_init+0x485/0x4e0
  ...
  Sep 14 12:02:53 kernel: [Firmware Bug]: cpu 0, try to use APIC520 (LVT offset 2) for vector 0xf4, but the register is already in use for vector 0x0 on this cpu

Without reported_ignored_msrs=0 at least the host kernel log will contain
enough information to avoid going on a wild goose chase.  But if reports
about individual MSR accesses are being silenced too, at least complain
loudly the first time a VM is started.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 months agoKVM: VMX: don't include '<linux/find.h>' directly
Wolfram Sang [Tue, 17 Dec 2024 07:05:40 +0000 (08:05 +0100)]
KVM: VMX: don't include '<linux/find.h>' directly

The header clearly states that it does not want to be included directly,
only via '<linux/bitmap.h>'. Replace the include accordingly.

Signed-off-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Message-ID: <20241217070539.2433-2-wsa+renesas@sang-engineering.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
5 months agoMerge tag 'devicetree-fixes-for-6.13-1' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 22 Dec 2024 16:40:23 +0000 (08:40 -0800)]
Merge tag 'devicetree-fixes-for-6.13-1' of git://git./linux/kernel/git/robh/linux

Pull devicetree fixes from Rob Herring:

 - Disable #address-cells/#size-cells warning on coreboot (Chromebooks)
   platforms

 - Add missing root #address-cells/#size-cells in default empty DT

 - Fix uninitialized variable in of_irq_parse_one()

 - Fix interrupt-map cell length check in of_irq_parse_imap_parent()

 - Fix refcount handling in __of_get_dma_parent()

 - Fix error path in of_parse_phandle_with_args_map()

 - Fix dma-ranges handling with flags cells

 - Drop explicit fw_devlink handling of 'interrupt-parent'

 - Fix "compression" typo in fixed-partitions binding

 - Unify "fsl,liodn" property type definitions

* tag 'devicetree-fixes-for-6.13-1' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux:
  of: Add coreboot firmware to excluded default cells list
  of/irq: Fix using uninitialized variable @addr_len in API of_irq_parse_one()
  of/irq: Fix interrupt-map cell length check in of_irq_parse_imap_parent()
  of: Fix refcount leakage for OF node returned by __of_get_dma_parent()
  of: Fix error path in of_parse_phandle_with_args_map()
  dt-bindings: mtd: fixed-partitions: Fix "compression" typo
  of: Add #address-cells/#size-cells in the device-tree root empty node
  dt-bindings: Unify "fsl,liodn" type definitions
  of: address: Preserve the flags portion on 1:1 dma-ranges mapping
  of/unittest: Add empty dma-ranges address translation tests
  of: property: fw_devlink: Do not use interrupt-parent directly

5 months agoMerge tag 'soc-fixes-6.13-2' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Linus Torvalds [Sat, 21 Dec 2024 23:45:06 +0000 (15:45 -0800)]
Merge tag 'soc-fixes-6.13-2' of git://git./linux/kernel/git/soc/soc

Pull SoC fixes from Arnd Bergmann:
 "Two more small fixes, correcting the cacheline size on Raspberry Pi 5
  and fixing a logic mistake in the microchip mpfs firmware driver"

* tag 'soc-fixes-6.13-2' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc:
  arm64: dts: broadcom: Fix L2 linesize for Raspberry Pi 5
  firmware: microchip: fix UL_IAP lock check in mpfs_auto_update_state()

5 months agoMerge tag 'mm-hotfixes-stable-2024-12-21-12-09' of git://git.kernel.org/pub/scm/linux...
Linus Torvalds [Sat, 21 Dec 2024 23:31:56 +0000 (15:31 -0800)]
Merge tag 'mm-hotfixes-stable-2024-12-21-12-09' of git://git./linux/kernel/git/akpm/mm

Pull misc fixes from Andrew Morton:
 "25 hotfixes.  16 are cc:stable.  19 are MM and 6 are non-MM.

  The usual bunch of singletons and doubletons - please see the relevant
  changelogs for details"

* tag 'mm-hotfixes-stable-2024-12-21-12-09' of git://git.kernel.org/pub/scm/linux/kernel/git/akpm/mm: (25 commits)
  mm: huge_memory: handle strsep not finding delimiter
  alloc_tag: fix set_codetag_empty() when !CONFIG_MEM_ALLOC_PROFILING_DEBUG
  alloc_tag: fix module allocation tags populated area calculation
  mm/codetag: clear tags before swap
  mm/vmstat: fix a W=1 clang compiler warning
  mm: convert partially_mapped set/clear operations to be atomic
  nilfs2: fix buffer head leaks in calls to truncate_inode_pages()
  vmalloc: fix accounting with i915
  mm/page_alloc: don't call pfn_to_page() on possibly non-existent PFN in split_large_buddy()
  fork: avoid inappropriate uprobe access to invalid mm
  nilfs2: prevent use of deleted inode
  zram: fix uninitialized ZRAM not releasing backing device
  zram: refuse to use zero sized block device as backing device
  mm: use clear_user_(high)page() for arch with special user folio handling
  mm: introduce cpu_icache_is_aliasing() across all architectures
  mm: add RCU annotation to pte_offset_map(_lock)
  mm: correctly reference merged VMA
  mm: use aligned address in copy_user_gigantic_page()
  mm: use aligned address in clear_gigantic_page()
  mm: shmem: fix ShmemHugePages at swapout
  ...

5 months agostaging: gpib: Fix allyesconfig build failures
Steven Rostedt [Tue, 17 Dec 2024 15:19:04 +0000 (10:19 -0500)]
staging: gpib: Fix allyesconfig build failures

My tests run an allyesconfig build and it failed with the following errors:

    LD [M]  samples/kfifo/dma-example.ko
  ld.lld: error: undefined symbol: nec7210_board_reset
  ld.lld: error: undefined symbol: nec7210_read
  ld.lld: error: undefined symbol: nec7210_write

It appears that some modules call the function nec7210_board_reset()
that is defined in nec7210.c.  In an allyesconfig build, these other
modules are built in.  But the file that holds nec7210_board_reset()
has:

  obj-m += nec7210.o

Where that "-m" means it only gets built as a module. With the other
modules built in, they have no access to nec7210_board_reset() and the build
fails.

This isn't the only function. After fixing that one, I hit another:

  ld.lld: error: undefined symbol: push_gpib_event
  ld.lld: error: undefined symbol: gpib_match_device_path

Where push_gpib_event() was also used outside of the file it was defined
in, and that file too only was built as a module.

Since the directory that nec7210.c is only traversed when
CONFIG_GPIB_NEC7210 is set, and the directory with gpib_common.c is only
traversed when CONFIG_GPIB_COMMON is set, use those configs as the
option to build those modules.  When it is an allyesconfig, then they
will both be built in and their functions will be available to the other
modules that are also built in.

Fixes: 3ba84ac69b53e ("staging: gpib: Add nec7210 GPIB chip driver")
Fixes: 9dde4559e9395 ("staging: gpib: Add GPIB common core driver")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 months agoMerge tag 'kbuild-fixes-v6.13-2' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 21 Dec 2024 19:24:32 +0000 (11:24 -0800)]
Merge tag 'kbuild-fixes-v6.13-2' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull Kbuild fixes from Masahiro Yamada:

 - Remove stale code in usr/include/headers_check.pl

 - Fix issues in the user-mode-linux Debian package

 - Fix false-positive "export twice" errors in modpost

* tag 'kbuild-fixes-v6.13-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  modpost: distinguish same module paths from different dump files
  kbuild: deb-pkg: Do not install maint scripts for arch 'um'
  kbuild: deb-pkg: add debarch for ARCH=um
  kbuild: Drop support for include/asm-<arch> in headers_check.pl

5 months agoMerge tag 'bpf-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
Linus Torvalds [Sat, 21 Dec 2024 19:07:19 +0000 (11:07 -0800)]
Merge tag 'bpf-fixes' of git://git./linux/kernel/git/bpf/bpf

Pull BPF fixes from Daniel Borkmann:

 - Fix inlining of bpf_get_smp_processor_id helper for !CONFIG_SMP
   systems (Andrea Righi)

 - Fix BPF USDT selftests helper code to use asm constraint "m" for
   LoongArch (Tiezhu Yang)

 - Fix BPF selftest compilation error in get_uprobe_offset when
   PROCMAP_QUERY is not defined (Jerome Marchand)

 - Fix BPF bpf_skb_change_tail helper when used in context of BPF
   sockmap to handle negative skb header offsets (Cong Wang)

 - Several fixes to BPF sockmap code, among others, in the area of
   socket buffer accounting (Levi Zim, Zijian Zhang, Cong Wang)

* tag 'bpf-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  selftests/bpf: Test bpf_skb_change_tail() in TC ingress
  selftests/bpf: Introduce socket_helpers.h for TC tests
  selftests/bpf: Add a BPF selftest for bpf_skb_change_tail()
  bpf: Check negative offsets in __bpf_skb_min_len()
  tcp_bpf: Fix copied value in tcp_bpf_sendmsg
  skmsg: Return copied bytes in sk_msg_memcopy_from_iter
  tcp_bpf: Add sk_rmem_alloc related logic for tcp_bpf ingress redirection
  tcp_bpf: Charge receive socket buffer in bpf_tcp_ingress()
  selftests/bpf: Fix compilation error in get_uprobe_offset()
  selftests/bpf: Use asm constraint "m" for LoongArch
  bpf: Fix bpf_get_smp_processor_id() on !CONFIG_SMP

5 months agoMerge tag 'media/v6.13-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Sat, 21 Dec 2024 18:56:34 +0000 (10:56 -0800)]
Merge tag 'media/v6.13-3' of git://git./linux/kernel/git/mchehab/linux-media

Pull media fixes from Mauro Carvalho Chehab:

 - fix a clang build issue with mediatec vcodec

 - add missing variable initialization to dib3000mb write function

* tag 'media/v6.13-3' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  media: mediatek: vcodec: mark vdec_vp9_slice_map_counts_eob_coef noinline
  media: dvb-frontends: dib3000mb: fix uninit-value in dib3000_write_reg

5 months agoMerge tag 'pci-v6.13-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/pci/pci
Linus Torvalds [Sat, 21 Dec 2024 18:51:04 +0000 (10:51 -0800)]
Merge tag 'pci-v6.13-fixes-2' of git://git./linux/kernel/git/pci/pci

Pull PCI fixes from Krzysztof Wilczyński:
 "Two small patches that are important for fixing boot time hang on
  Intel JHL7540 'Titan Ridge' platforms equipped with a Thunderbolt
  controller.

  The boot time issue manifests itself when a PCI Express bandwidth
  control is unnecessarily enabled on the Thunderbolt controller
  downstream ports, which only supports a link speed of 2.5 GT/s in
  accordance with USB4 v2 specification (p. 671, sec. 11.2.1, "PCIe
  Physical Layer Logical Sub-block").

  As such, there is no need to enable bandwidth control on such
  downstream port links, which also works around the issue.

  Both patches were tested by the original reporter on the hardware on
  which the failure origin golly manifested itself. Both fixes were
  proven to resolve the reported boot hang issue, and both patches have
  been in linux-next this week with no reported problems"

* tag 'pci-v6.13-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/pci/pci:
  PCI/bwctrl: Enable only if more than one speed is supported
  PCI: Honor Max Link Speed when determining supported speeds

5 months agoMerge tag 'pm-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm
Linus Torvalds [Sat, 21 Dec 2024 18:47:47 +0000 (10:47 -0800)]
Merge tag 'pm-6.13-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull power management fixes from Rafael Wysocki:
 "These fix some amd-pstate driver issues:

   - Detect preferred core support in amd-pstate before driver
     registration to avoid initialization ordering issues (K Prateek
     Nayak)

   - Fix issues with with boost numerator handling in amd-pstate leading
     to inconsistently programmed CPPC max performance values (Mario
     Limonciello)"

* tag 'pm-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  cpufreq/amd-pstate: Use boost numerator for upper bound of frequencies
  cpufreq/amd-pstate: Store the boost numerator as highest perf again
  cpufreq/amd-pstate: Detect preferred core support before driver registration

5 months agoMerge tag 'thermal-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 21 Dec 2024 18:44:44 +0000 (10:44 -0800)]
Merge tag 'thermal-6.13-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull thermal control fixes from Rafael Wysocki:
 "Fix two issues with the user thermal thresholds feature introduced in
  this development cycle (Daniel Lezcano)"

* tag 'thermal-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  thermal/thresholds: Fix boundaries and detection routine
  thermal/thresholds: Fix uapi header macros leading to a compilation error

5 months agoMerge tag 'acpi-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael...
Linus Torvalds [Sat, 21 Dec 2024 18:42:35 +0000 (10:42 -0800)]
Merge tag 'acpi-6.13-rc4' of git://git./linux/kernel/git/rafael/linux-pm

Pull ACPI fix from Rafael Wysocki:
 "Unbreak ACPI EC support on LoongArch that has been broken earlier in
  this development cycle (Huacai Chen)"

* tag 'acpi-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  ACPI: EC: Enable EC support on LoongArch by default

5 months agoMerge tag '6.13-rc3-SMB3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sat, 21 Dec 2024 17:35:18 +0000 (09:35 -0800)]
Merge tag '6.13-rc3-SMB3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6

Pull smb client fixes from Steve French:

 - fix regression in display of write stats

 - fix rmmod failure with network namespaces

 - two minor cleanups

* tag '6.13-rc3-SMB3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
  smb: fix bytes written value in /proc/fs/cifs/Stats
  smb: client: fix TCP timers deadlock after rmmod
  smb: client: Deduplicate "select NETFS_SUPPORT" in Kconfig
  smb: use macros instead of constants for leasekey size and default cifsattrs value

5 months agoMerge tag 'nfs-for-6.13-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Sat, 21 Dec 2024 17:32:24 +0000 (09:32 -0800)]
Merge tag 'nfs-for-6.13-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client fixes from Trond Myklebust:

 - NFS/pnfs: Fix a live lock between recalled layouts and layoutget

 - Fix a build warning about an undeclared symbol 'nfs_idmap_cache_timeout'

* tag 'nfs-for-6.13-2' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  fs/nfs: fix missing declaration of nfs_idmap_cache_timeout
  NFS/pnfs: Fix a live lock between recalled layouts and layoutget

5 months agoMerge tag 'ceph-for-6.13-rc4' of https://github.com/ceph/ceph-client
Linus Torvalds [Sat, 21 Dec 2024 17:29:46 +0000 (09:29 -0800)]
Merge tag 'ceph-for-6.13-rc4' of https://github.com/ceph/ceph-client

Pull ceph fixes from Ilya Dryomov:
 "A handful of important CephFS fixes from Max, Alex and myself: memory
  corruption due to a buffer overrun, potential infinite loop and
  several memory leaks on the error paths. All but one marked for
  stable"

* tag 'ceph-for-6.13-rc4' of https://github.com/ceph/ceph-client:
  ceph: allocate sparse_ext map only for sparse reads
  ceph: fix memory leak in ceph_direct_read_write()
  ceph: improve error handling and short/overflow-read logic in __ceph_sync_read()
  ceph: validate snapdirname option length when mounting
  ceph: give up on paths longer than PATH_MAX
  ceph: fix memory leaks in __ceph_sync_read()

5 months agomodpost: distinguish same module paths from different dump files
Masahiro Yamada [Thu, 12 Dec 2024 15:46:15 +0000 (00:46 +0900)]
modpost: distinguish same module paths from different dump files

Since commit 13b25489b6f8 ("kbuild: change working directory to external
module directory with M="), module paths are always relative to the top
of the external module tree.

The module paths recorded in Module.symvers are no longer globally unique
when they are passed via KBUILD_EXTRA_SYMBOLS for building other external
modules, which may result in false-positive "exported twice" errors.
Such errors should not occur because external modules should be able to
override in-tree modules.

To address this, record the dump file path in struct module and check it
when searching for a module.

Fixes: 13b25489b6f8 ("kbuild: change working directory to external module directory with M=")
Reported-by: Jon Hunter <jonathanh@nvidia.com>
Closes: https://lore.kernel.org/all/eb21a546-a19c-40df-b821-bbba80f19a3d@nvidia.com/
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
5 months agokbuild: deb-pkg: Do not install maint scripts for arch 'um'
Nicolas Schier [Thu, 12 Dec 2024 13:05:29 +0000 (14:05 +0100)]
kbuild: deb-pkg: Do not install maint scripts for arch 'um'

Stop installing Debian maintainer scripts when building a
user-mode-linux Debian package.

Debian maintainer scripts are used for e.g. requesting rebuilds of
initrd, rebuilding DKMS modules and updating of grub configuration.  As
all of this is not relevant for UML but also may lead to failures while
processing the kernel hooks, do no more install maintainer scripts for
the UML package.

Suggested-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Nicolas Schier <nicolas@fjasle.eu>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
5 months agokbuild: deb-pkg: add debarch for ARCH=um
Masahiro Yamada [Tue, 3 Dec 2024 11:14:45 +0000 (20:14 +0900)]
kbuild: deb-pkg: add debarch for ARCH=um

'make ARCH=um bindeb-pkg' shows the following warning.

  $ make ARCH=um bindeb-pkg
     [snip]
    GEN     debian

  ** ** **  WARNING  ** ** **

  Your architecture doesn't have its equivalent
  Debian userspace architecture defined!
  Falling back to the current host architecture (amd64).
  Please add support for um to ./scripts/package/mkdebian ...

This commit hard-codes i386/amd64 because UML is only supported for x86.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
5 months agokbuild: Drop support for include/asm-<arch> in headers_check.pl
Geert Uytterhoeven [Thu, 5 Dec 2024 13:20:43 +0000 (14:20 +0100)]
kbuild: Drop support for include/asm-<arch> in headers_check.pl

"include/asm-<arch>" was replaced by "arch/<arch>/include/asm" a long
time ago.  All assembler header files are now included using
"#include <asm/*>", so there is no longer a need to rewrite paths.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
5 months agoselftests/bpf: Test bpf_skb_change_tail() in TC ingress
Cong Wang [Fri, 13 Dec 2024 03:40:57 +0000 (19:40 -0800)]
selftests/bpf: Test bpf_skb_change_tail() in TC ingress

Similarly to the previous test, we also need a test case to cover
positive offsets as well, TC is an excellent hook for this.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Zijian Zhang <zijianzhang@bytedance.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241213034057.246437-5-xiyou.wangcong@gmail.com
5 months agoselftests/bpf: Introduce socket_helpers.h for TC tests
Cong Wang [Fri, 13 Dec 2024 03:40:56 +0000 (19:40 -0800)]
selftests/bpf: Introduce socket_helpers.h for TC tests

Pull socket helpers out of sockmap_helpers.h so that they can be reused
for TC tests as well. This prepares for the next patch.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241213034057.246437-4-xiyou.wangcong@gmail.com
5 months agoselftests/bpf: Add a BPF selftest for bpf_skb_change_tail()
Cong Wang [Fri, 13 Dec 2024 03:40:55 +0000 (19:40 -0800)]
selftests/bpf: Add a BPF selftest for bpf_skb_change_tail()

As requested by Daniel, we need to add a selftest to cover
bpf_skb_change_tail() cases in skb_verdict. Here we test trimming,
growing and error cases, and validate its expected return values and the
expected sizes of the payload.

Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241213034057.246437-3-xiyou.wangcong@gmail.com
5 months agobpf: Check negative offsets in __bpf_skb_min_len()
Cong Wang [Fri, 13 Dec 2024 03:40:54 +0000 (19:40 -0800)]
bpf: Check negative offsets in __bpf_skb_min_len()

skb_network_offset() and skb_transport_offset() can be negative when
they are called after we pull the transport header, for example, when
we use eBPF sockmap at the point of ->sk_data_ready().

__bpf_skb_min_len() uses an unsigned int to get these offsets, this
leads to a very large number which then causes bpf_skb_change_tail()
failed unexpectedly.

Fix this by using a signed int to get these offsets and ensure the
minimum is at least zero.

Fixes: 5293efe62df8 ("bpf: add bpf_skb_change_tail helper")
Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241213034057.246437-2-xiyou.wangcong@gmail.com
5 months agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Fri, 20 Dec 2024 22:10:01 +0000 (14:10 -0800)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fix from Catalin Marinas:
 "Fix a sparse warning in the arm64 signal code dealing with the user
  shadow stack register, GCSPR_EL0"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  arm64/signal: Silence sparse warning storing GCSPR_EL0

5 months agotcp_bpf: Fix copied value in tcp_bpf_sendmsg
Levi Zim [Sat, 30 Nov 2024 13:38:23 +0000 (21:38 +0800)]
tcp_bpf: Fix copied value in tcp_bpf_sendmsg

bpf kselftest sockhash::test_txmsg_cork_hangs in test_sockmap.c triggers a
kernel NULL pointer dereference:

BUG: kernel NULL pointer dereference, address: 0000000000000008
 ? __die_body+0x6e/0xb0
 ? __die+0x8b/0xa0
 ? page_fault_oops+0x358/0x3c0
 ? local_clock+0x19/0x30
 ? lock_release+0x11b/0x440
 ? kernelmode_fixup_or_oops+0x54/0x60
 ? __bad_area_nosemaphore+0x4f/0x210
 ? mmap_read_unlock+0x13/0x30
 ? bad_area_nosemaphore+0x16/0x20
 ? do_user_addr_fault+0x6fd/0x740
 ? prb_read_valid+0x1d/0x30
 ? exc_page_fault+0x55/0xd0
 ? asm_exc_page_fault+0x2b/0x30
 ? splice_to_socket+0x52e/0x630
 ? shmem_file_splice_read+0x2b1/0x310
 direct_splice_actor+0x47/0x70
 splice_direct_to_actor+0x133/0x300
 ? do_splice_direct+0x90/0x90
 do_splice_direct+0x64/0x90
 ? __ia32_sys_tee+0x30/0x30
 do_sendfile+0x214/0x300
 __se_sys_sendfile64+0x8e/0xb0
 __x64_sys_sendfile64+0x25/0x30
 x64_sys_call+0xb82/0x2840
 do_syscall_64+0x75/0x110
 entry_SYSCALL_64_after_hwframe+0x4b/0x53

This is caused by tcp_bpf_sendmsg() returning a larger value(12289) than
size (8192), which causes the while loop in splice_to_socket() to release
an uninitialized pipe buf.

The underlying cause is that this code assumes sk_msg_memcopy_from_iter()
will copy all bytes upon success but it actually might only copy part of
it.

This commit changes it to use the real copied bytes.

Signed-off-by: Levi Zim <rsworktech@outlook.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Björn Töpel <bjorn@kernel.org>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241130-tcp-bpf-sendmsg-v1-2-bae583d014f3@outlook.com
5 months agoskmsg: Return copied bytes in sk_msg_memcopy_from_iter
Levi Zim [Sat, 30 Nov 2024 13:38:22 +0000 (21:38 +0800)]
skmsg: Return copied bytes in sk_msg_memcopy_from_iter

Previously sk_msg_memcopy_from_iter returns the copied bytes from the
last copy_from_iter{,_nocache} call upon success.

This commit changes it to return the total number of copied bytes on
success.

Signed-off-by: Levi Zim <rsworktech@outlook.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Björn Töpel <bjorn@kernel.org>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241130-tcp-bpf-sendmsg-v1-1-bae583d014f3@outlook.com
5 months agoMerge tag 'hwmon-for-v6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 20 Dec 2024 21:48:41 +0000 (13:48 -0800)]
Merge tag 'hwmon-for-v6.13-rc4' of git://git./linux/kernel/git/groeck/linux-staging

Pull hwmon fixes from Guenter Roeck:

 - Fix reporting of negative temperature, current, and voltage values in
   the tmp513 driver

* tag 'hwmon-for-v6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/linux-staging:
  hwmon: (tmp513) Fix interpretation of values of Temperature Result and Limit Registers
  hwmon: (tmp513) Fix Current Register value interpretation
  hwmon: (tmp513) Fix interpretation of values of Shunt Voltage and Limit Registers

5 months agoof: Add coreboot firmware to excluded default cells list
Rob Herring (Arm) [Fri, 20 Dec 2024 21:06:47 +0000 (15:06 -0600)]
of: Add coreboot firmware to excluded default cells list

Google Juniper and other Chromebook platforms have a very old bootloader
which populates /firmware node without proper address/size-cells leading
to warnings:

  Missing '#address-cells' in /firmware
  WARNING: CPU: 0 PID: 1 at drivers/of/base.c:106 of_bus_n_addr_cells+0x90/0xf0
  Modules linked in:
  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Not tainted 6.12.0 #1 933ab9971ff4d5dc58cb378a96f64c7f72e3454d
  Hardware name: Google juniper sku16 board (DT)
  ...
  Missing '#size-cells' in /firmware
  WARNING: CPU: 0 PID: 1 at drivers/of/base.c:133 of_bus_n_size_cells+0x90/0xf0
  Modules linked in:
  CPU: 0 UID: 0 PID: 1 Comm: swapper/0 Tainted: G        W          6.12.0 #1 933ab9971ff4d5dc58cb378a96f64c7f72e3454d
  Tainted: [W]=WARN
  Hardware name: Google juniper sku16 board (DT)

These platform won't receive updated bootloader/firmware, so add an
exclusion for platforms with a "coreboot" compatible node. While this is
wider than necessary, that's the easiest fix and it doesn't doesn't
matter if we miss checking other platforms using coreboot.

We may revisit this later and address with a fixup to the DT itself.

Reported-by: Sasha Levin <sashal@kernel.org>
Closes: https://lore.kernel.org/all/Z0NUdoG17EwuCigT@sashalap/
Cc: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: Matthias Brugger <matthias.bgg@gmail.com>
Cc: Chen-Yu Tsai <wenst@chromium.org>
Cc: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Rob Herring (Arm) <robh@kernel.org>
5 months agoMerge tag 'block-6.13-20241220' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 20 Dec 2024 21:37:58 +0000 (13:37 -0800)]
Merge tag 'block-6.13-20241220' of git://git.kernel.dk/linux

Pull block fixes from Jens Axboe:

 - Minor cleanups for bdev/nvme using the helpers introduced

 - Revert of a deadlock fix that still needs more work

 - Fix a UAF of hctx in the cpu hotplug code

* tag 'block-6.13-20241220' of git://git.kernel.dk/linux:
  block: avoid to reuse `hctx` not removed from cpuhp callback list
  block: Revert "block: Fix potential deadlock while freezing queue and acquiring sysfs_lock"
  nvme: use blk_validate_block_size() for max LBA check
  block/bdev: use helper for max block size check

5 months agoMerge tag 'io_uring-6.13-20241220' of git://git.kernel.dk/linux
Linus Torvalds [Fri, 20 Dec 2024 21:32:43 +0000 (13:32 -0800)]
Merge tag 'io_uring-6.13-20241220' of git://git.kernel.dk/linux

Pull io_uring fixes from Jens Axboe:

 - Fix for a file ref leak for registered ring fds

 - Turn the ->timeout_lock into a raw spinlock, as it nests under the
   io-wq lock which is a raw spinlock as it's called from the scheduler
   side

 - Limit ring resizing to DEFER_TASKRUN for now. We will broaden this in
   the future, but for now, ensure that it's only feasible on rings with
   a single user

 - Add sanity check for io-wq enqueuing

* tag 'io_uring-6.13-20241220' of git://git.kernel.dk/linux:
  io_uring: check if iowq is killed before queuing
  io_uring/register: limit ring resizing to DEFER_TASKRUN
  io_uring: Fix registered ring file refcount leak
  io_uring: make ctx->timeout_lock a raw spinlock

5 months agoMerge tag 'usb-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Fri, 20 Dec 2024 19:09:40 +0000 (11:09 -0800)]
Merge tag 'usb-6.13-rc4' of git://git./linux/kernel/git/gregkh/usb

Pull USB / Thunderbolt fixes from Greg KH:
 "Here are some important, and small, fixes for USB and Thunderbolt
  issues that have come up in the -rc releases. And some new device ids
  for good measure. Included in here are:

   - Much reported xhci bugfix for usb-storage devices (and other
     devices as well, tripped me up on a video camera)

   - thunderbolt fixes for some small reported issues

   - new usb-serial device ids

  All of these have been in linux-next this week with no reported issues"

* tag 'usb-6.13-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: xhci: fix ring expansion regression in 6.13-rc1
  xhci: Turn NEC specific quirk for handling Stop Endpoint errors generic
  thunderbolt: Improve redrive mode handling
  USB: serial: option: add Telit FE910C04 rmnet compositions
  USB: serial: option: add MediaTek T7XX compositions
  USB: serial: option: add Netprisma LCUK54 modules for WWAN Ready
  USB: serial: option: add MeiG Smart SLM770A
  USB: serial: option: add TCL IK512 MBIM & ECM
  thunderbolt: Don't display nvm_version unless upgrade supported
  thunderbolt: Add support for Intel Panther Lake-M/P

5 months agoMerge tag 'spi-fix-v6.13-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/brooni...
Linus Torvalds [Fri, 20 Dec 2024 19:06:25 +0000 (11:06 -0800)]
Merge tag 'spi-fix-v6.13-rc3' of git://git./linux/kernel/git/broonie/spi

Pull spi fix from Mark Brown:
 "A fix for the remove path of the Rockchip driver, the code was just
  clearly and obviously wrong"

* tag 'spi-fix-v6.13-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
  spi: rockchip-sfc: Fix error in remove progress

5 months agoMerge tag 'regulator-fix-v6.13-rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 20 Dec 2024 19:04:02 +0000 (11:04 -0800)]
Merge tag 'regulator-fix-v6.13-rc3' of git://git./linux/kernel/git/broonie/regulator

Pull regulator fix from Mark Brown:
 "The recently added regulator-uv-survival-time-ms property was renamed
  during the review of the series that added it, but unfortunately only
  in the DT binding and not in the code that parses the binding.

  This brings the code in line with the binding, if someone started
  using the original name we can add compat support for it but there's
  nothing upstream yet and it's a very niche feature so hopefully not"

* tag 'regulator-fix-v6.13-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: rename regulator-uv-survival-time-ms according to DT binding

5 months agoMerge tag 'drm-fixes-2024-12-20' of https://gitlab.freedesktop.org/drm/kernel
Linus Torvalds [Fri, 20 Dec 2024 18:17:53 +0000 (10:17 -0800)]
Merge tag 'drm-fixes-2024-12-20' of https://gitlab.freedesktop.org/drm/kernel

Pull drm fixes from Dave Airlie:
 "Probably the last pull before Christmas holidays, I'll still be around
  for most of the time anyways, nothing too major in here, bunch of
  amdgpu and i915 along with a smattering of fixes across the board.

  core:
   - fix FB dependency
   - avoid div by 0 more in vrefresh
   - maintainers update

  display:
   - fix DP tunnel error path

  dma-buf:
   - fix !DEBUG_FS

  sched:
   - docs warning fix

  panel:
   - collection of misc panel fixes

  i915:
   - Reset engine utilization buffer before registration
   - Ensure busyness counter increases motonically
   - Accumulate active runtime on gt reset

  amdgpu:
   - Disable BOCO when CONFIG_HOTPLUG_PCI_PCIE is not enabled
   - scheduler job fixes
   - IP version check fixes
   - devcoredump fix
   - GPUVM update fix
   - NBIO 2.5 fix

  udmabuf:
   - fix memory leak on last export
   - sealing fixes

  ivpu:
   - fix NULL pointer
   - fix memory leak
   - fix WARN"

* tag 'drm-fixes-2024-12-20' of https://gitlab.freedesktop.org/drm/kernel: (33 commits)
  drm/sched: Fix drm_sched_fini() docu generation
  accel/ivpu: Fix WARN in ivpu_ipc_send_receive_internal()
  accel/ivpu: Fix memory leak in ivpu_mmu_reserved_context_init()
  accel/ivpu: Fix general protection fault in ivpu_bo_list()
  drm/amdgpu/nbio7.0: fix IP version check
  drm/amd: Update strapping for NBIO 2.5.0
  drm/amdgpu: Handle NULL bo->tbo.resource (again) in amdgpu_vm_bo_update
  drm/amdgpu: fix amdgpu_coredump
  drm/amdgpu/smu14.0.2: fix IP version check
  drm/amdgpu/gfx12: fix IP version check
  drm/amdgpu/mmhub4.1: fix IP version check
  drm/amdgpu/nbio7.11: fix IP version check
  drm/amdgpu/nbio7.7: fix IP version check
  drm/amdgpu: don't access invalid sched
  drm/amd: Require CONFIG_HOTPLUG_PCI_PCIE for BOCO
  drm: rework FB_CORE dependency
  drm/fbdev: Select FB_CORE dependency for fbdev on DMA and TTM
  fbdev: Fix recursive dependencies wrt BACKLIGHT_CLASS_DEVICE
  i915/guc: Accumulate active runtime on gt reset
  i915/guc: Ensure busyness counter increases motonically
  ...

5 months agoMerge tag 'trace-ringbuffer-v6.13-rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 20 Dec 2024 18:13:26 +0000 (10:13 -0800)]
Merge tag 'trace-ringbuffer-v6.13-rc3' of git://git./linux/kernel/git/trace/linux-trace

Pull ring-buffer fixes from Steven Rostedt:

 - Fix possible overflow of mmapped ring buffer with bad offset

   If the mmap() to the ring buffer passes in a start address that is
   passed the end of the mmapped file, it is not caught and a
   slab-out-of-bounds is triggered.

   Add a check to make sure the start address is within the bounds

 - Do not use TP_printk() to boot mapped ring buffers

   As a boot mapped ring buffer's data may have pointers that map to the
   previous boot's memory map, it is unsafe to allow the TP_printk() to
   be used to read the boot mapped buffer's events. If a TP_printk()
   points to a static string from within the kernel it will not match
   the current kernel mapping if KASLR is active, and it can fault.

   Have it simply print out the raw fields.

* tag 'trace-ringbuffer-v6.13-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  trace/ring-buffer: Do not use TP_printk() formatting for boot mapped buffers
  ring-buffer: Fix overflow in __rb_map_vma

5 months agoMerge tag 'arm-soc/for-6.13/devicetree-arm64-fixes' of https://github.com/Broadcom...
Arnd Bergmann [Fri, 20 Dec 2024 17:02:27 +0000 (18:02 +0100)]
Merge tag 'arm-soc/for-6.13/devicetree-arm64-fixes' of https://github.com/Broadcom/stblinux into arm/fixes

This pull request contains Broadcom ARM64-based SoCs Device Tree fixes
for 6.13, please pull the following:

- Willow corrects the L2 cache line size on the Raspberry Pi 5 (2712) to
  the correct value of 64 bytes

* tag 'arm-soc/for-6.13/devicetree-arm64-fixes' of https://github.com/Broadcom/stblinux:
  arm64: dts: broadcom: Fix L2 linesize for Raspberry Pi 5

Link: https://lore.kernel.org/r/20241217190547.868744-1-florian.fainelli@broadcom.com
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
5 months agoMerge tag 'riscv-soc-fixes-for-v6.13-rc4' of https://git.kernel.org/pub/scm/linux...
Arnd Bergmann [Fri, 20 Dec 2024 17:00:30 +0000 (18:00 +0100)]
Merge tag 'riscv-soc-fixes-for-v6.13-rc4' of https://git./linux/kernel/git/conor/linux into arm/fixes

RISC-V soc driver fixes for v6.13-rc4

A single fix for the Auto Update driver, where a mistake in array
indexing (accessing as a u32 rather than a u8) caused the driver to read
the wrong feature disable bits.

Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
* tag 'riscv-soc-fixes-for-v6.13-rc4' of https://git.kernel.org/pub/scm/linux/kernel/git/conor/linux:
  firmware: microchip: fix UL_IAP lock check in mpfs_auto_update_state()

Link: https://lore.kernel.org/r/20241218-suffrage-unfazed-fa0113072a42@spud
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
5 months agotcp_bpf: Add sk_rmem_alloc related logic for tcp_bpf ingress redirection
Zijian Zhang [Tue, 10 Dec 2024 01:20:39 +0000 (01:20 +0000)]
tcp_bpf: Add sk_rmem_alloc related logic for tcp_bpf ingress redirection

When we do sk_psock_verdict_apply->sk_psock_skb_ingress, an sk_msg will
be created out of the skb, and the rmem accounting of the sk_msg will be
handled by the skb.

For skmsgs in __SK_REDIRECT case of tcp_bpf_send_verdict, when redirecting
to the ingress of a socket, although we sk_rmem_schedule and add sk_msg to
the ingress_msg of sk_redir, we do not update sk_rmem_alloc. As a result,
except for the global memory limit, the rmem of sk_redir is nearly
unlimited. Thus, add sk_rmem_alloc related logic to limit the recv buffer.

Since the function sk_msg_recvmsg and __sk_psock_purge_ingress_msg are
used in these two paths. We use "msg->skb" to test whether the sk_msg is
skb backed up. If it's not, we shall do the memory accounting explicitly.

Fixes: 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: Zijian Zhang <zijianzhang@bytedance.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241210012039.1669389-3-zijianzhang@bytedance.com
5 months agotcp_bpf: Charge receive socket buffer in bpf_tcp_ingress()
Cong Wang [Tue, 10 Dec 2024 01:20:38 +0000 (01:20 +0000)]
tcp_bpf: Charge receive socket buffer in bpf_tcp_ingress()

When bpf_tcp_ingress() is called, the skmsg is being redirected to the
ingress of the destination socket. Therefore, we should charge its
receive socket buffer, instead of sending socket buffer.

Because sk_rmem_schedule() tests pfmemalloc of skb, we need to
introduce a wrapper and call it for skmsg.

Fixes: 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: Cong Wang <cong.wang@bytedance.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20241210012039.1669389-2-zijianzhang@bytedance.com
5 months agoarm64/signal: Silence sparse warning storing GCSPR_EL0
Mark Brown [Sat, 14 Dec 2024 02:12:06 +0000 (02:12 +0000)]
arm64/signal: Silence sparse warning storing GCSPR_EL0

We are seeing a sparse warning in gcs_restore_signal():

  arch/arm64/kernel/signal.c:1054:9: sparse: sparse: cast removes address space '__user' of expression

when storing the final GCSPR_EL0 value back into the register, caused by
the fact that write_sysreg_s() casts the value it writes to a u64 which
sparse sees as discarding the __userness of the pointer.

Avoid this by treating the address as an integer, casting to a pointer only
when using it to write to userspace.

While we're at it also inline gcs_signal_cap_valid() into it's one user
and make equivalent updates to gcs_signal_entry().

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202412082005.OBJ0BbWs-lkp@intel.com/
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20241214-arm64-gcs-signal-sparse-v3-1-5e8d18fffc0c@kernel.org
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
5 months agoMerge tag 'amd-drm-fixes-6.13-2024-12-18' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 20 Dec 2024 02:53:42 +0000 (12:53 +1000)]
Merge tag 'amd-drm-fixes-6.13-2024-12-18' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-6.13-2024-12-18:

amdgpu:
- Disable BOCO when CONFIG_HOTPLUG_PCI_PCIE is not enabled
- scheduler job fixes
- IP version check fixes
- devcoredump fix
- GPUVM update fix
- NBIO 2.5 fix

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20241218204637.2966198-1-alexander.deucher@amd.com
5 months agoKVM: x86/mmu: Treat TDP MMU faults as spurious if access is already allowed
Sean Christopherson [Wed, 18 Dec 2024 21:36:11 +0000 (13:36 -0800)]
KVM: x86/mmu: Treat TDP MMU faults as spurious if access is already allowed

Treat slow-path TDP MMU faults as spurious if the access is allowed given
the existing SPTE to fix a benign warning (other than the WARN itself)
due to replacing a writable SPTE with a read-only SPTE, and to avoid the
unnecessary LOCK CMPXCHG and subsequent TLB flush.

If a read fault races with a write fault, fast GUP fails for any reason
when trying to "promote" the read fault to a writable mapping, and KVM
resolves the write fault first, then KVM will end up trying to install a
read-only SPTE (for a !map_writable fault) overtop a writable SPTE.

Note, it's not entirely clear why fast GUP fails, or if that's even how
KVM ends up with a !map_writable fault with a writable SPTE.  If something
else is going awry, e.g. due to a bug in mmu_notifiers, then treating read
faults as spurious in this scenario could effectively mask the underlying
problem.

However, retrying the faulting access instead of overwriting an existing
SPTE is functionally correct and desirable irrespective of the WARN, and
fast GUP _can_ legitimately fail with a writable VMA, e.g. if the Accessed
bit in primary MMU's PTE is toggled and causes a PTE value mismatch.  The
WARN was also recently added, specifically to track down scenarios where
KVM is unnecessarily overwrites SPTEs, i.e. treating the fault as spurious
doesn't regress KVM's bug-finding capabilities in any way.  In short,
letting the WARN linger because there's a tiny chance it's due to a bug
elsewhere would be excessively paranoid.

Fixes: 1a175082b190 ("KVM: x86/mmu: WARN and flush if resolving a TDP MMU fault clears MMU-writable")
Reported-by: Lei Yang <leiyang@redhat.com>
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=219588
Tested-by: Lei Yang <leiyang@redhat.com>
Link: https://lore.kernel.org/r/20241218213611.3181643-1-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>
5 months agoKVM: SVM: Allow guest writes to set MSR_AMD64_DE_CFG bits
Sean Christopherson [Wed, 11 Dec 2024 17:29:52 +0000 (09:29 -0800)]
KVM: SVM: Allow guest writes to set MSR_AMD64_DE_CFG bits

Drop KVM's arbitrary behavior of making DE_CFG.LFENCE_SERIALIZE read-only
for the guest, as rejecting writes can lead to guest crashes, e.g. Windows
in particular doesn't gracefully handle unexpected #GPs on the WRMSR, and
nothing in the AMD manuals suggests that LFENCE_SERIALIZE is read-only _if
it exists_.

KVM only allows LFENCE_SERIALIZE to be set, by the guest or host, if the
underlying CPU has X86_FEATURE_LFENCE_RDTSC, i.e. if LFENCE is guaranteed
to be serializing.  So if the guest sets LFENCE_SERIALIZE, KVM will provide
the desired/correct behavior without any additional action (the guest's
value is never stuffed into hardware).  And having LFENCE be serializing
even when it's not _required_ to be is a-ok from a functional perspective.

Fixes: 74a0e79df68a ("KVM: SVM: Disallow guest from changing userspace's MSR_AMD64_DE_CFG value")
Fixes: d1d93fa90f1a ("KVM: SVM: Add MSR-based feature support for serializing LFENCE")
Reported-by: Simon Pilkington <simonp.git@mailbox.org>
Closes: https://lore.kernel.org/all/52914da7-a97b-45ad-86a0-affdf8266c61@mailbox.org
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: stable@vger.kernel.org
Reviewed-by: Tom Lendacky <thomas.lendacky@amd.com>
Link: https://lore.kernel.org/r/20241211172952.1477605-1-seanjc@google.com
Signed-off-by: Sean Christopherson <seanjc@google.com>