Modifying the size of the global bin_attribute instance can be racy.
Instead use the new .bin_size callback to do so safely.
For this to work move the initialization of c2dev->ops before the call
to device_create() as the size callback will need access to it.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://lore.kernel.org/r/20241221-sysfs-const-bin_attr-misc-drivers-v2-4-ba5e79fe8771@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The sysfs core now allows instances of 'struct bin_attribute' 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>
Acked-by: Andrew Donnellan <ajd@linux.ibm.com>
Link: https://lore.kernel.org/r/20241221-sysfs-const-bin_attr-misc-drivers-v2-2-ba5e79fe8771@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The sysfs core now allows instances of 'struct bin_attribute' 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>
Acked-by: Andrew Donnellan <ajd@linux.ibm.com>
Link: https://lore.kernel.org/r/20241221-sysfs-const-bin_attr-misc-drivers-v2-1-ba5e79fe8771@weissschuh.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use per-vma locking in the shrinker's callback when reclaiming pages,
similar to the page installation logic. This minimizes contention with
unrelated vmas improving performance. The mmap_sem is still acquired if
the per-vma lock cannot be obtained.
Cc: Suren Baghdasaryan <surenb@google.com>
Suggested-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-10-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of always overriding errors with -ENOMEM, propagate the specific
error code returned by vm_insert_page(). This allows for more accurate
error logs and handling.
Cc: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-9-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Use per-vma locking for concurrent page installations, this minimizes
contention with unrelated vmas improving performance. The mmap_lock is
still acquired when needed though, e.g. before get_user_pages_remote().
Many thanks to Barry Song who posted a similar approach [1].
Link: https://lore.kernel.org/all/20240902225009.34576-1-21cnbao@gmail.com/ [1]
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Hillf Danton <hdanton@sina.com>
Cc: Lorenzo Stoakes <lorenzo.stoakes@oracle.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-8-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The alloc->buffer field in struct binder_alloc stores the starting
address of the mapped vma, rename this field to alloc->vm_start to
better reflect its purpose. It also avoids confusion with the binder
buffer concept, e.g. transaction->buffer.
No functional changes in this patch.
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-7-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It is unsafe to use alloc->vma outside of the mmap_sem. Instead, add a
new boolean alloc->mapped to save the vma state (mapped or unmmaped) and
use this as a replacement for alloc->vma to validate several paths.
Using the alloc->vma caused several performance and security issues in
the past. Now that it has been replaced with either vm_lookup() or the
alloc->mapped state, we can finally remove it.
Cc: Minchan Kim <minchan@kernel.org>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-6-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of pre-allocating an entire array of struct binder_lru_page in
alloc->pages, install the shrinker metadata under page->private. This
ensures the memory is allocated and released as needed alongside pages.
By converting the alloc->pages[] into an array of struct page pointers,
we can access these pages directly and only reference the shrinker
metadata where it's being used (e.g. inside the shrinker's callback).
Rename struct binder_lru_page to struct binder_shrinker_mdata to better
reflect its purpose. Add convenience functions that wrap the allocation
and freeing of pages along with their shrinker metadata.
Note I've reworked this patch to avoid using page->lru and page->index
directly, as Matthew pointed out that these are being removed [1].
Link: https://lore.kernel.org/all/ZzziucEm3np6e7a0@casper.infradead.org/ [1]
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-5-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The numa node id for binder pages is currently being derived from the
lru entry under struct binder_lru_page. However, this object doesn't
reflect the node id of the struct page items allocated separately.
Instead, select the correct node id from the page itself. This was made
possible since commit 0a97c01cd2 ("list_lru: allow explicit memcg and
NUMA node selection").
Cc: Nhat Pham <nphamcs@gmail.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Suren Baghdasaryan <surenb@google.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-4-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Allow multiple callers to install pages simultaneously by switching the
mmap_sem from write-mode to read-mode. Races to the same PTE are handled
using get_user_pages_remote() to retrieve the already installed page.
This method significantly reduces contention in the mmap semaphore.
To ensure safety, vma_lookup() is used (instead of alloc->vma) to avoid
operating on an isolated VMA. In addition, zap_page_range_single() is
called under the alloc->mutex to avoid racing with the shrinker.
Many thanks to Barry Song who posted a similar approach [1].
Link: https://lore.kernel.org/all/20240902225009.34576-1-21cnbao@gmail.com/ [1]
Cc: David Hildenbrand <david@redhat.com>
Cc: Barry Song <v-songbaohua@oppo.com>
Cc: Suren Baghdasaryan <surenb@google.com>
Cc: Liam R. Howlett <Liam.Howlett@oracle.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-3-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This reverts commit 7710e2cca3.
In preparation for concurrent page installations, restore the original
alloc->mutex which will serialize zap_page_range_single() against page
installations in subsequent patches (instead of the mmap_sem).
Resolved trivial conflicts with commit 2c10a20f5e ("binder_alloc: Fix
sleeping function called from invalid context") and commit da0c02516c
("mm/list_lru: simplify the list_lru walk callback function").
Cc: Mukesh Ojha <quic_mojha@quicinc.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Carlos Llamas <cmllamas@google.com>
Link: https://lore.kernel.org/r/20241210143114.661252-2-cmllamas@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Set TARGET_SCF_ACK_KREF flag and allow f_tcm to take the BOT command
reference. A usb request may be canceled, the f_tcm knows this. Let it
decides if the command should be freed. This is the same as how the UAS
interface is done.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/e791c639e91b5d91a8787f5d6902e8c58f1dc172.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Handle target_core_fabric_ops TASK MANAGEMENT functions and their
response. If a TASK MANAGEMENT command is received, the driver will
interpret the function TMF_*, translate to TMR_*, and fire off a command
work executing target_submit_tmr(). On completion, it will handle the
TASK MANAGEMENT response through uasp_send_tm_response().
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/50339586e36509dadb9c208b3314530993e673b6.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
If the transfer is cancelled due to a disconnect or driver tear down
(error code -ESHUTDOWN), then just free the command. However, if it got
cancelled due to other reasons, then send a sense CHECK CONDITION status
with TCM_CHECK_CONDITION_ABORT_CMD status to host notifying the delivery
failure. Note that this is separate from TASK MANAGEMENT function abort
task command, which will require a separate response IU.
See UAS-r04 section 8.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/f2ae293c1fc39df4d242a2f724584bf4ec105ece.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There may be different reasons for the transfer to be cancelled. Don't
blindly free the command without checking its status. We may still need
to properly respond to cancelled command. Check and only free the
command on endpoint disable.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/6c4ae2b4c2a9037bdcb6f909e173a94b11f04657.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Don't just wait for the data write completion and execute the target
command. We need to verify if the request completed successfully and not
just sending invalid data. The verification is done in the write request
completion routine. Queue the same work of the command to execute the
target_execute_cmd() on data write.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/9f6b1c6946cf49eeba0173e405678b9b7786636b.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
To properly respond to host sending more commands than the number of
streams the device advertises, the device needs to be able to reject the
command with a response. Allocate an extra request to handle 1 more
command than the number of streams.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/256f2ec8f5e042ab692d9593144fa75f3d3ce94b.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Resubmit command on completion to fetch more commands and service them
in parallel. Increase the number of work in a queue. Each work will be
for each command allowing them to be processed concurrently. Also, set
them to be unbounded by cpu to improve performance.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/933cf7191b672bf4cfbea4df19af1b08dc1baca9.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
We can handle multiple commands concurently. Each command services a
stream id. At the moment, the driver will handle 32 outstanding streams,
which is equivalent to 32 commands. Make sure to allocate a matching
number of commands to the number of streams.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/2d806120dcc10c88fef21865b7bc1d2b6604fe42.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Host can assign stream ID value greater than number of streams
allocated. The tcm function needs to keep track of which stream is
available to assign the stream ID. This patch doesn't track that, but at
least it makes sure that there's no Oops if the host send tag with a
value greater than the number of supported streams.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/d57b7dfd228199cef811206b1b83ec649f742752.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Only allocate up to UASP_SS_EP_COMP_NUM_STREAMS number of session tags.
We should not be using more than UASP_SS_EP_COMP_NUM_STREAMS of tags due
to the number of commands limit we imposed. Each command uses a unique
tag. Any more than that is unnecessary. By limiting it, we can detect an
issue in our driver immediately should we run out of session tags.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/017016ffcab2f3c284d863fc42483b83dbd21b35.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently the default bMaxBurst is 0. Set default bMaxBurst to 15 (i.e.
16 bursts) to Data IN and OUT endpoints to improve performance. It
should be fine for a controller that supports less than 16 bursts. It
should be able to negotiate properly with the host at packet level for
the end of burst.
If the controller can't handle a burst of 16, and high performance isn't
important, the user can use BOT protocol from mass_storage gadget driver
instead.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/9d6265db4d138e542f281988362bc4392b034d43.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some old builds of Microsoft Windows 10 UASP class driver reject UASP
device with stream count of 2^4. To keep compatibility with both Linux
and Windows, let's increase the stream count to 2^5. Also, internal
tests show that stream count of 2^5 increases performance slightly.
Signed-off-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/23bf7f5cb04da691fd6ba0a77babee9ad3195f44.1733876548.git.Thinh.Nguyen@synopsys.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Altmodes with cros_ec are either automatically entered by the EC or
entered by the AP if TBT or USB4 are supported on the system. Due to the
security risk of PCIe tunneling, TBT modes should not be auto entered by
the kernel at this time and will require user intervention.
With this change, a userspace program will need to explicitly activate
the thunderbolt mode on the port and partner in order to enter the mode
and the thunderbolt driver will not automatically enter when a partner
is connected.
Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Benson Leung <bleung@chromium.org>
Link: https://lore.kernel.org/r/20241213153543.v5.8.Ic14738918e3d026fa2d85e95fb68f8e07a0828d0@changeid
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Mux configuration is often the final piece of mode entry and can be used
to determine whether a partner altmode is active. When mux configuration
is done, use the active port altmode's SVID to set the partner active
field for all partner alt modes.
Signed-off-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Reviewed-by: Benson Leung <bleung@chromium.org>
Link: https://lore.kernel.org/r/20241213153543.v5.5.I083bf9188947be8cb7460211cfdf3233370a28f6@changeid
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
