shmem_destroy isn't always required for coprocessor-managed buffers but we
still enforce that it exists. Just relax the check.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Reviewed-by: Neal Gompa <neal@gompa.dev>
Signed-off-by: Sven Peter <sven@kernel.org>
Link: https://lore.kernel.org/r/20250610-smc-6-15-v7-4-556cafd771d3@kernel.org
Signed-off-by: Lee Jones <lee@kernel.org>
Certain messages from DCP contain NUL bytes in the random data after the
NUL terminated syslog message. Since the syslog message ends with '\n'
this results in a dev_info() message terminated with two newlines and an
empty printed line in the kernel log.
Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Link: https://lore.kernel.org/r/20250226-apple-soc-misc-v2-4-c3ec37f9021b@svenpeter.dev
Signed-off-by: Sven Peter <sven@svenpeter.dev>
rtkit messages as communication with the DCP firmware for framebuffer
swaps or input events are time critical so use WQ_HIGHPRI to prevent
user space CPU load to increase latency.
With kwin_wayland 6's explicit sync mode user space load was able to
delay the IOMFB rtkit communication enough to miss vsync for surface
swaps. Minimal test scenario is constantly resizing a glxgears
Xwayland window.
Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Link: https://lore.kernel.org/r/20250226-apple-soc-misc-v2-3-c3ec37f9021b@svenpeter.dev
Signed-off-by: Sven Peter <sven@svenpeter.dev>
Apparently nobody can figure out where the old logic came from, but it
seems like it has never been actually used on any supported firmware to
this day. OSLog buffers were apparently never requested.
But starting with 13.3, we actually need this implemented properly for
MTP (and later AOP) to work, so let's actually do that.
Signed-off-by: Hector Martin <marcan@marcan.st>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Link: https://lore.kernel.org/r/20250226-apple-soc-misc-v2-2-c3ec37f9021b@svenpeter.dev
Signed-off-by: Sven Peter <sven@svenpeter.dev>
This state is needed to wake the dcp IOP after m1n1 shut it down
and works for all other co-processors as well.
Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Link: https://lore.kernel.org/r/20250226-apple-soc-misc-v2-1-c3ec37f9021b@svenpeter.dev
Signed-off-by: Sven Peter <sven@svenpeter.dev>
This code calls kfree(bfr); and then passes "bfr" to rtk->ops->crashed()
which is a use after free. The ->crashed function pointer is implemented
by apple_nvme_rtkit_crashed() and it doesn't use the "bfr" pointer so
this doesn't cause a problem. But it still looks sketchy as can be.
Fix this by moving kfree() after the last usage of bfr.
Fixes: bf8b4e4977 ("soc: apple: rtkit: Pass the crashlog to the crashed() callback")
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Reviewed-by: Eric Curtin <ecurtin@redhat.com>
Link: https://lore.kernel.org/r/20250212085853.1357906-1-harshit.m.mogalapalli@oracle.com
Signed-off-by: Sven Peter <sven@svenpeter.dev>
Client drivers might want a copy of the crashlog to stash into a
devcoredump blob. Since device memory management can be very variable,
the actual devcoredump implementation is left to client drivers. Pass
the raw crashlog buffer to the client callback so it can use it if
desired.
Signed-off-by: Asahi Lina <lina@asahilina.net>
Reviewed-by: Jens Axboe <axboe@kernel.dk>
Link: https://lore.kernel.org/r/20250202-rtkit-crashdump-v1-1-9d38615b4e12@asahilina.net
Signed-off-by: Sven Peter <sven@svenpeter.dev>
Now that we have a mailbox driver in drivers/soc/apple, port the RTKit
code to it. This mostly just entails replacing calls through the mailbox
subsystem with direct calls into the driver.
Acked-by: Eric Curtin <ecurtin@redhat.com>
Acked-by: Neal Gompa <neal@gompa.dev>
Acked-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Signed-off-by: Hector Martin <marcan@marcan.st>
It is fundamentally broken and has no users. Just remove it.
Acked-by: Eric Curtin <ecurtin@redhat.com>
Acked-by: Neal Gompa <neal@gompa.dev>
Acked-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Signed-off-by: Hector Martin <marcan@marcan.st>
Crop trailing whitespace, null, and newline characters in syslog
messages received from coprocessors. Notably DCP sends its messages
including a trailing newline, so prior to this change we would end up
cluttering the kernel log by repeated newlines at the end of messages.
Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
Reviewed-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Hector Martin <marcan@marcan.st>
The buffer address field is missing two bits. This matters for the GPU,
which uses upper-half 64-bit addresses on the ASC and those get sign
extended from the mailbox message field, so the right number of high
bits need to be set.
Signed-off-by: Asahi Lina <lina@asahilina.net>
Reviewed-by: Sven Peter <sven@svenpeter.dev>
Signed-off-by: Hector Martin <marcan@marcan.st>
While we normally encourage devm usage by drivers, some consumers (and
in particular the upcoming Rust abstractions) might want to manually
manage memory. Export the raw functions to make this possible.
Signed-off-by: Asahi Lina <lina@asahilina.net>
Reviewed-by: Sven Peter <sven@svenpeter.dev>
Reviewed-by: Eric Curtin <ecurtin@redhat.com>
Signed-off-by: Hector Martin <marcan@marcan.st>
This is yet another low power mode, used by DCP.
Reviewed-by: Eric Curtin <ecurtin@redhat.com>
Reviewed-by: Sven Peter <sven@svenpeter.dev>
Signed-off-by: Hector Martin <marcan@marcan.st>
This allows a client to receive messages in atomic context, by polling.
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Sven Peter <sven@svenpeter.dev>
Reviewed-by: Eric Curtin <ecurtin@redhat.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Apple SoCs such as the M1 come with multiple embedded co-processors
running proprietary firmware. Communication with those is established
over a simple mailbox using the RTKit IPC protocol.
This cannot be implemented inside the mailbox subsystem since on top
of communication over channels we also need support for starting,
hibernating and resetting these co-processors. We also need to
handle shared memory allocations differently depending on the
co-processor and don't want to split that across multiple drivers.
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sven Peter <sven@svenpeter.dev>
