Magnus Karlsson says:
====================
net: xdp: execute xdp_do_flush() before napi_complete_done()
Make sure that xdp_do_flush() is always executed before
napi_complete_done(). This is important for two reasons. First, a
redirect to an XSKMAP assumes that a call to xdp_do_redirect() from
napi context X on CPU Y will be followed by a xdp_do_flush() from the
same napi context and CPU. This is not guaranteed if the
napi_complete_done() is executed before xdp_do_flush(), as it tells
the napi logic that it is fine to schedule napi context X on another
CPU. Details from a production system triggering this bug using the
veth driver can be found in [1].
The second reason is that the XDP_REDIRECT logic in itself relies on
being inside a single NAPI instance through to the xdp_do_flush() call
for RCU protection of all in-kernel data structures. Details can be
found in [2].
The drivers have only been compile-tested since I do not own any of
the HW below. So if you are a maintainer, it would be great if you
could take a quick look to make sure I did not mess something up.
Note that these were the drivers I found that violated the ordering by
running a simple script and manually checking the ones that came up as
potential offenders. But the script was not perfect in any way. There
might still be offenders out there, since the script can generate
false negatives.
[1] https://lore.kernel.org/r/20221220185903.1105011-1-sbohrer@cloudflare.com
[2] https://lore.kernel.org/all/20210624160609.292325-1-toke@redhat.com/
====================
Link: https://lore.kernel.org/r/20230125074901.2737-1-magnus.karlsson@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Make sure that xdp_do_flush() is always executed before
napi_complete_done(). This is important for two reasons. First, a
redirect to an XSKMAP assumes that a call to xdp_do_redirect() from
napi context X on CPU Y will be followed by a xdp_do_flush() from the
same napi context and CPU. This is not guaranteed if the
napi_complete_done() is executed before xdp_do_flush(), as it tells
the napi logic that it is fine to schedule napi context X on another
CPU. Details from a production system triggering this bug using the
veth driver can be found following the first link below.
The second reason is that the XDP_REDIRECT logic in itself relies on
being inside a single NAPI instance through to the xdp_do_flush() call
for RCU protection of all in-kernel data structures. Details can be
found in the second link below.
Fixes: d678be1dc1 ("dpaa2-eth: add XDP_REDIRECT support")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://lore.kernel.org/r/20221220185903.1105011-1-sbohrer@cloudflare.com
Link: https://lore.kernel.org/all/20210624160609.292325-1-toke@redhat.com/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Make sure that xdp_do_flush() is always executed before
napi_complete_done(). This is important for two reasons. First, a
redirect to an XSKMAP assumes that a call to xdp_do_redirect() from
napi context X on CPU Y will be followed by a xdp_do_flush() from the
same napi context and CPU. This is not guaranteed if the
napi_complete_done() is executed before xdp_do_flush(), as it tells
the napi logic that it is fine to schedule napi context X on another
CPU. Details from a production system triggering this bug using the
veth driver can be found following the first link below.
The second reason is that the XDP_REDIRECT logic in itself relies on
being inside a single NAPI instance through to the xdp_do_flush() call
for RCU protection of all in-kernel data structures. Details can be
found in the second link below.
Fixes: a1e031ffb4 ("dpaa_eth: add XDP_REDIRECT support")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://lore.kernel.org/r/20221220185903.1105011-1-sbohrer@cloudflare.com
Link: https://lore.kernel.org/all/20210624160609.292325-1-toke@redhat.com/
Acked-by: Camelia Groza <camelia.groza@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Make sure that xdp_do_flush() is always executed before
napi_complete_done(). This is important for two reasons. First, a
redirect to an XSKMAP assumes that a call to xdp_do_redirect() from
napi context X on CPU Y will be followed by a xdp_do_flush() from the
same napi context and CPU. This is not guaranteed if the
napi_complete_done() is executed before xdp_do_flush(), as it tells
the napi logic that it is fine to schedule napi context X on another
CPU. Details from a production system triggering this bug using the
veth driver can be found following the first link below.
The second reason is that the XDP_REDIRECT logic in itself relies on
being inside a single NAPI instance through to the xdp_do_flush() call
for RCU protection of all in-kernel data structures. Details can be
found in the second link below.
Fixes: 186b3c998c ("virtio-net: support XDP_REDIRECT")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://lore.kernel.org/r/20221220185903.1105011-1-sbohrer@cloudflare.com
Link: https://lore.kernel.org/all/20210624160609.292325-1-toke@redhat.com/
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Make sure that xdp_do_flush() is always executed before
napi_complete_done(). This is important for two reasons. First, a
redirect to an XSKMAP assumes that a call to xdp_do_redirect() from
napi context X on CPU Y will be followed by a xdp_do_flush() from the
same napi context and CPU. This is not guaranteed if the
napi_complete_done() is executed before xdp_do_flush(), as it tells
the napi logic that it is fine to schedule napi context X on another
CPU. Details from a production system triggering this bug using the
veth driver can be found following the first link below.
The second reason is that the XDP_REDIRECT logic in itself relies on
being inside a single NAPI instance through to the xdp_do_flush() call
for RCU protection of all in-kernel data structures. Details can be
found in the second link below.
Fixes: a825b611c7 ("net: lan966x: Add support for XDP_REDIRECT")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Steen Hegelund <Steen.Hegelund@microchip.com>
Link: https://lore.kernel.org/r/20221220185903.1105011-1-sbohrer@cloudflare.com
Link: https://lore.kernel.org/all/20210624160609.292325-1-toke@redhat.com/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Make sure that xdp_do_flush() is always executed before
napi_complete_done(). This is important for two reasons. First, a
redirect to an XSKMAP assumes that a call to xdp_do_redirect() from
napi context X on CPU Y will be followed by a xdp_do_flush() from the
same napi context and CPU. This is not guaranteed if the
napi_complete_done() is executed before xdp_do_flush(), as it tells
the napi logic that it is fine to schedule napi context X on another
CPU. Details from a production system triggering this bug using the
veth driver can be found following the first link below.
The second reason is that the XDP_REDIRECT logic in itself relies on
being inside a single NAPI instance through to the xdp_do_flush() call
for RCU protection of all in-kernel data structures. Details can be
found in the second link below.
Fixes: d1b25b79e1 ("qede: add .ndo_xdp_xmit() and XDP_REDIRECT support")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://lore.kernel.org/r/20221220185903.1105011-1-sbohrer@cloudflare.com
Link: https://lore.kernel.org/all/20210624160609.292325-1-toke@redhat.com/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
-----BEGIN PGP SIGNATURE-----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=/W/P
-----END PGP SIGNATURE-----
Merge tag '6.2-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6
Pull cifs fix from Steve French:
"Fix for reconnect oops in smbdirect (RDMA), also is marked for stable"
* tag '6.2-rc5-smb3-client-fixes' of git://git.samba.org/sfrench/cifs-2.6:
cifs: Fix oops due to uncleared server->smbd_conn in reconnect
-----BEGIN PGP SIGNATURE-----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=rJIx
-----END PGP SIGNATURE-----
Merge tag 'io_uring-6.2-2023-01-27' of git://git.kernel.dk/linux
Pull io_uring fixes from Jens Axboe:
"Two small fixes for this release:
- Sanitize how async prep is done for drain requests, so we ensure
that it always gets done (Dylan)
- A ring provided buffer recycling fix for multishot receive (me)"
* tag 'io_uring-6.2-2023-01-27' of git://git.kernel.dk/linux:
io_uring: always prep_async for drain requests
io_uring/net: cache provided buffer group value for multishot receives
Instead of waiting for an RCU grace period between each ipc_namespace
structure that is being freed, wait an RCU grace period for every batch
of ipc_namespace structures.
Thanks to Al Viro for the suggestion of the helper function.
This speeds up the run time of the test case that allocates ipc_namespaces
in a loop from 6 minutes, to a little over 1 second:
real 0m1.192s
user 0m0.038s
sys 0m1.152s
Signed-off-by: Rik van Riel <riel@surriel.com>
Reported-by: Chris Mason <clm@meta.com>
Tested-by: Giuseppe Scrivano <gscrivan@redhat.com>
Suggested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Currently the ipc namespace allocation will fail when there are
ipc_namespace structures pending to be freed. This results in the
simple test case below, as well as some real world workloads, to
get allocation failures even when the number of ipc namespaces in
actual use is way below the limit.
int main()
{
int i;
for (i = 0; i < 100000; i++) {
if (unshare(CLONE_NEWIPC) < 0)
error(EXIT_FAILURE, errno, "unshare");
}
}
Make the allocation of an ipc_namespace wait for pending frees,
so it will succeed.
real 6m19.197s
user 0m0.041s
sys 0m1.019s
Signed-off-by: Rik van Riel <riel@surriel.com>
Reported-by: Chris Mason <clm@meta.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
- Fix filter memory leak by calling ftrace_free_filter()
- Initialize trace_printk() earlier so that ftrace_dump_on_oops shows data
on early crashes.
- Update the outdated instructions in scripts/tracing/ftrace-bisect.sh
- Add lockdep_is_held() to fix lockdep warning
- Add allocation failure check in create_hist_field()
- Don't initialize pointer that gets set right away in enabled_monitors_write()
- Update MAINTAINER entries
- Fix help messages in Kconfigs
- Fix kernel-doc header for update_preds()
-----BEGIN PGP SIGNATURE-----
iIoEABYIADIWIQRRSw7ePDh/lE+zeZMp5XQQmuv6qgUCY9MaCBQccm9zdGVkdEBn
b29kbWlzLm9yZwAKCRAp5XQQmuv6qqb4AP4p78mTVP/Rw5oLCRUM4E7JGb8hmxuZ
HD0RR9fbvyDaQAD/ffkar/KmvGCfCvBNrkGvvx98dwtGIssIB1dShoEZPAU=
=lBwh
-----END PGP SIGNATURE-----
Merge tag 'trace-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace
Pull tracing fixes from Steven Rostedt:
- Fix filter memory leak by calling ftrace_free_filter()
- Initialize trace_printk() earlier so that ftrace_dump_on_oops shows
data on early crashes.
- Update the outdated instructions in scripts/tracing/ftrace-bisect.sh
- Add lockdep_is_held() to fix lockdep warning
- Add allocation failure check in create_hist_field()
- Don't initialize pointer that gets set right away in enabled_monitors_write()
- Update MAINTAINER entries
- Fix help messages in Kconfigs
- Fix kernel-doc header for update_preds()
* tag 'trace-v6.2-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
bootconfig: Update MAINTAINERS file to add tree and mailing list
rv: remove redundant initialization of pointer ptr
ftrace: Maintain samples/ftrace
tracing/filter: fix kernel-doc warnings
lib: Kconfig: fix spellos
trace_events_hist: add check for return value of 'create_hist_field'
tracing/osnoise: Use built-in RCU list checking
tracing: Kconfig: Fix spelling/grammar/punctuation
ftrace/scripts: Update the instructions for ftrace-bisect.sh
tracing: Make sure trace_printk() can output as soon as it can be used
ftrace: Export ftrace_free_filter() to modules
Add VID and PID to the xpad_device table to allow driver
to use the 8BitDo Pro 2 Wired Controller, which is
XTYPE_XBOX360 compatible by default.
Signed-off-by: John Butler <radon86dev@gmail.com>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
Link: https://lore.kernel.org/r/20230124005206.80706-1-radon86dev@gmail.com
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
The previous change was not properly build tested and needs
a trivial spelling change:
ipaq-micro-ts.c:146:8: error: type defaults to 'int' in declaration of 'DEFINE_SIMPLE_DEV_PM' [-Werror=implicit-int]
Fixes: 144ff5e03d ("Input: ipaq-micro-ts - use DEFINE_SIMPLE_DEV_PM_OPS() and pm_sleep_ptr()")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230117164539.1631336-1-arnd@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
As the guards only apply to suspend and resume, #ifdef CONFIG_PM_SLEEP
would have been a tighter protection. As pm_sleep_ptr() lets the compiler
see the protected ops structure and callbacks but also lets the compiler
remove it as unused code if !CONFIG_PM_SLEEP this allows the #ifdef
guards to be removed, slightly simplifying the resulting code.
--
It seems likely that DEFINE_SIMPLE_DEV_PM_OPS() would work here but
I'd prefer not to make that change unless someone can confirm that the
extra callbacks registered will have no unwanted side effects in this
driver.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-17-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
As the guards only apply to suspend and resume, #ifdef CONFIG_PM_SLEEP
would have been a tighter protection. As pm_sleep_ptr() lets the compiler
see the protected ops structure and callbacks but also lets the compiler
remove it as unused code if !CONFIG_PM_SLEEP this allows the #ifdef
guards to be removed, slightly simplifying the resulting code.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
--
It seems likely that DEFINE_SIMPLE_DEV_PM_OPS() would work here but
I'd prefer not to make that change unless someone can confirm that the
extra callbacks registered will have no unwanted side effects in this
driver.
Link: https://lore.kernel.org/r/20230114171620.42891-16-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
As the new pm_sleep_ptr() macro lets the compiler see the code, but
then remove it if !CONFIG_PM_SLEEP it can be used to avoid the need
for #ifdef guards. Use that in the input core to simplify the code
a little. Note pm_sleep_ptr() has not been applied to each callback
in the ops structure because the pm_sleep_ptr() at the usage site
is sufficient.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-15-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_RUNTIME_DEV_PM_OPS() is deprecated as it requires explicit protection
against unused function warnings. The new combination of pm_ptr()
and RUNTIME_DEV_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to be
removed. Thus also drop the __maybe_unused markings.
Note that DEFINE_RUNTIME_DEV_PM_OPS() is not used because that adds
additional callbacks for suspend and resume and would need
testing.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-14-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() is deprecated as it requires explicit protection
against unused function warnings. The new combination of pm_sleep_ptr()
and SYSTEM_SLEEP_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to be
removed. Thus also drop the __maybe_unused markings.
In this case we also have a .poweroff_late() callback. Whilst not
strictly necessary, to future proof against relaxation of the protection
of the main driver.pm = pm_sleep_ptr() protect this pointer with
pm_sleep_ptr() as would be done if the LATE_SYSTEM_SLEEP_PM_OPS()
macro were used to set it.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-13-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SIMPLE_DEV_PM_OPS() is deprecated as it requires explicit protection
against unused function warnings. The new combination of pm_sleep_ptr()
and EXPORT_GPL_SIMPLE_DEV_PMU_OPS() allows the compiler to see the
functions, thus suppressing the warning, but still allowing the unused
code to be removed. Thus also drop the __maybe_unused markings.
It also rolls in the EXPORT_SYMBOL() so that we only export it
if CONFIG_PM_SLEEP.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Javier Martinez Canillas <javier@dowhile0.org>
Cc: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20230114171620.42891-12-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() and RUNTIME_PM_OPS() are deprecated as
they requires explicit protection against unused function warnings.
The new combination of pm_ptr() EXPORT_GPL_RUNTIME_DEV_PM_OPS()
allows the compiler to see the functions, thus suppressing the
warning, but still allowing the unused code to be
removed. Thus also drop the #ifdef guards.
Note that we are replacing an unconditional call to the suspend
and resume functions for sleep use cases with one via
pm_runtime_force_suspend() / pm_runtime_force_resume() that only
do anything to the device if we are not already in the appropriate
runtime suspended state.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
--
I 'think' this should be fine in that it can only reduce the number
of unnecessary suspends. If anyone can test that would be great.
Link: https://lore.kernel.org/r/20230114171620.42891-11-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SIMPLE_DEV_PM_OPS() is deprecated as it requires explicit protection
against unused function warnings. The new combination of pm_sleep_ptr()
and EXPORT_GPL_SIMPLE_DEV_PM_OPS() allows the compiler to see the
functions, thus suppressing the warning, but still allowing the unused
code to be removed. Thus also drop the __maybe_unused markings.
This function also removes the need for separate EXPORT_SYMBOL_GPL()
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-10-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
The I2C and SPI PM callbacks were identical (though wrapped in some
bouncing out to the bus specific container of the struct device and
then back again to get the drvdata). As such rather than just moving
these to SIMPLE_DEV_PM_OPS() and pm_sleep_ptr() take the opportunity
to unify the struct dev_pm_ops and use the new EXPORT_SIMPLE_DEV_PM_OPS()
macro so that we can drop the unused suspend and resume callbacks as well
as the structure if !CONFIG_PM_SLEEP without needing to mark the callbacks
__maybe_unused.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Michael Hennerich <michael.hennerich@analog.com>
Link: https://lore.kernel.org/r/20230114171620.42891-9-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
The I2C and SPI PM callbacks were identical (though wrapped in some
bouncing out to the bus specific container of the struct device and
then back again to get the drvdata). As such rather than just moving
these to SIMPLE_DEV_PM_OPS() and pm_sleep_ptr() take the opportunity
to unify the struct dev_pm_ops and use the new EXPORT_SIMPLE_DEV_PM_OPS()
macro so that we can drop the unused suspend and resume callbacks as well
as the structure if !CONFIG_PM_SLEEP without needing to mark the callbacks
__maybe_unused.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Michael Hennerich <michael.hennerich@analog.com>
Link: https://lore.kernel.org/r/20230114171620.42891-8-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() and SET_RUNTIME_PM_OPS() are deprecated as
they require explicit protection against unused function warnings.
The new combination of pm_ptr() and SYSTEM_SLEEP_PM_OPS()/
RUNTIME_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to be
removed. Thus also drop the __maybe_unused markings.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Andi Shyti <andi@etezian.org>
Link: https://lore.kernel.org/r/20230114171620.42891-7-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() and SET_RUNTIME_PM_OPS() are deprecated
as they requires explicit protection against unused function warnings.
The new combination of pm_ptr() and SYSTEM_SLEEP_PM_OPS() /
RUNTIME_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to
be removed. Thus also drop the #ifdef guards.
Whilst all 3 sets of callbacks are similar, there are small differences
that make it challenging to use a single pm_dev_ops structure - hence
leave the duplication as it stands.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Matthias Schiffer <matthias.schiffer@ew.tq-group.com>
Cc: Lyude Paul <lyude@redhat.com>
Cc: Andrew Duggan <aduggan@synaptics.com>
Link: https://lore.kernel.org/r/20230114171620.42891-6-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() and SET_RUNTIME_PM_OPS() are deprecated as
they require explicit protection against unused function warnings.
The new combination of pm_ptr() and SYSTEM_SLEEP_PM_OPS()/
RUNTIME_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to be
removed. Thus also drop the __maybe_unused markings.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Caleb Connolly <caleb@connolly.tech>
Cc: Andi Shyti <andi@etezian.org>
Tested-by: Caleb Connolly <caleb@connolly.tech> # oneplus-guacamole
Link: https://lore.kernel.org/r/20230114171620.42891-5-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
The SET_ variants are deprecated as they require explicit protection
against unused function warnings. The new combination of pm_ptr()
and SYSTEM_SLEEP/RUNTIME_DEV_PM_OPS() allow the compiler to see the
functions, thus suppressing the warning, but still allowing the unused
code to be removed. Thus also drop the #ifdef guards.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-4-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() is deprecated as it requires explicit protection
against unused function warnings. The new combination of pm_sleep_ptr()
and SYSTEM_SLEEP_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to be
removed. Thus also drop the __maybe_unused markings. Here the
additional .resume_noirq callback is protected with pm_sleep_ptr(). This
isn't strictly necessary but is done for consistency with the other
callbacks.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Cc: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20230114171620.42891-3-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
SET_SYSTEM_SLEEP_PM_OPS() and SET_RUNTIME_PM_OPS() are deprecated as
they require explicit protection against unused function warnings.
The new combination of pm_ptr() and SYSTEM_SLEEP_PM_OPS()/
RUNTIME_PM_OPS() allows the compiler to see the functions,
thus suppressing the warning, but still allowing the unused code to be
removed. Thus also drop the __maybe_unused markings.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230114171620.42891-2-jic23@kernel.org
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Try to make fs/verity/verify.c aware of large folios. This includes
making fsverity_verify_bio() support the case where the bio contains
large folios, and adding a function fsverity_verify_folio() which is the
equivalent of fsverity_verify_page().
There's no way to actually test this with large folios yet, but I've
tested that this doesn't cause any regressions.
Signed-off-by: Eric Biggers <ebiggers@google.com>
Link: https://lore.kernel.org/r/20230127221529.299560-1-ebiggers@kernel.org
This fixes the build here locally on my 32-bit arm build.
Signed-off-by: Dave Airlie <airlied@redhat.com>
(cherry picked from commit f439a959dc)
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
The Allwinner D1 family of SoCs contain a PPU power domain controller
separate from the PRCM. It can power down the video engine and DSP, and
it contains special logic for hardware-assisted CPU idle.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Link: https://lore.kernel.org/r/20230126063419.15971-4-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
The PPU contains a series of identical MMIO register ranges, one for
each power domain. Each range contains control/status bits for a clock
gate, reset line, output gates, and a power switch. (The clock and reset
are separate from, and in addition to, the bits in the CCU.) It also
contains a hardware power sequence engine to control the other bits.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Link: https://lore.kernel.org/r/20230126063419.15971-3-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
The Allwinner D1 family of SoCs contain a PPU power domain controller
separate from the PRCM. It can power down the video engine and DSP, and
it contains special logic for hardware-assisted CPU idle. Other recent
Allwinner SoCs (e.g. TV303) have a PPU with a different set of domains.
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Link: https://lore.kernel.org/r/20230126063419.15971-2-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Now that several D1-based boards are supported, enable the platform in
our defconfig. Build in the drivers which are necessary to boot, such as
the pinctrl, MMC, RTC (which provides critical clocks), SPI (for flash),
and watchdog (which may be left enabled by the bootloader). Other common
onboard peripherals are enabled as modules.
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Link: https://lore.kernel.org/r/20230126045738.47903-12-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Allwinner manufactures the sunxi family of application processors. This
includes the "sun8i" series of ARMv7 SoCs, the "sun50i" series of ARMv8
SoCs, and now the "sun20i" series of 64-bit RISC-V SoCs.
The first SoC in the sun20i series is D1, containing a single T-HEAD
C906 core. D1s is a low-pin-count variant of D1 with co-packaged DRAM.
Most peripherals are shared across the entire chip family. In fact, the
ARMv7 T113 SoC is pin-compatible and almost entirely register-compatible
with the D1s.
This means many existing device drivers can be reused. To facilitate
this reuse, name the symbol ARCH_SUNXI, since that is what the existing
drivers have as their dependency.
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Link: https://lore.kernel.org/r/20230126045738.47903-11-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
This fixes the doxygen format documentation above the
user_ring_buffer__* APIs. There has to be a newline
before the @brief, otherwise doxygen won't render them
for libbpf.readthedocs.org.
Signed-off-by: Grant Seltzer <grantseltzer@gmail.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20230126024749.522278-1-grantseltzer@gmail.com
The 100ask Dongshan Nezha STU is a system-on-module that can be used
standalone or with a carrier board. The SoM provides gigabit Ethernet,
HDMI, a USB peripheral port, and WiFi/Bluetooth via an RTL8723DS chip.
The "DIY" carrier board exposes almost every pin from the D1 SoC to 0.1"
headers, but contains no digital circuitry, so it does not have its own
devicetree.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-10-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
The MangoPi MQ Pro is a tiny SBC with a layout compatible to the
Raspberry Pi Zero. It includes the Allwinner D1 SoC, 512M or 1G of DDR3,
and an RTL8723DS-based WiFi/Bluetooth module.
The board also exposes GPIO Port E via a connector on the end of the
board, which can support either a camera or an RMII Ethernet PHY. The
additional regulators supply that connector.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-9-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Sipeed manufactures a "Lichee RV" system-on-module, which provides a
minimal working system on its own, as well as a few carrier boards. The
"Dock" board provides audio, USB, and WiFi. The "86 Panel" additionally
provides 100M Ethernet and a built-in display panel.
The 86 Panel repurposes the USB ID and VBUS detection GPIOs for its RGB
panel interface, since the USB OTG port is inaccessible inside the case.
Co-developed-by: Jisheng Zhang <jszhang@kernel.org>
Signed-off-by: Jisheng Zhang <jszhang@kernel.org>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-8-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
"D1 Nezha" is Allwinner's first-party development board for the D1 SoC.
It was shipped with 512M, 1G, or 2G of DDR3. It supports onboard audio,
HDMI, gigabit Ethernet, WiFi and Bluetooth, USB 2.0 host and OTG ports,
plus low-speed I/O from the SoC and a GPIO expander chip.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Conor Dooley <conor.dooley@microchip.com>
Tested-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-7-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
The MangoPi MQ is a tiny SBC built around the Allwinner D1s. Its
onboard peripherals include two USB Type-C ports (1 device, 1 host)
and RTL8189FTV WLAN.
A MangoPi MQ-R variant of the board also exists. The MQ-R has a
different form factor, but the onboard peripherals are the same.
Most D1 and D1s boards use a similar power tree, with the 1.8V rail
powered by the SoC's internal LDOA, analog domains powered by ALDO,
and the rest of the board powered by always-on fixed regulators. To
avoid duplication, factor out the regulator information that is
common across boards.
The board also exposes GPIO Port E via a FPC connector, which can
support either a camera or an RMII Ethernet PHY. The additional
regulators supply that connector.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-6-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
D1 (aka D1-H), D1s (aka F133), R528, and T113 are a family of SoCs based
on a single die, or at a pair of dies derived from the same design.
D1 and D1s contain a single T-HEAD Xuantie C906 CPU, whereas R528 and
T113 contain a pair of Cortex-A7's. D1 and R528 are the full version of
the chip with a BGA package, whereas D1s and T113 are low-pin-count QFP
variants.
Because the original design supported both ARM and RISC-V CPUs, some
peripherals are duplicated. In addition, all variants except D1s contain
a HiFi 4 DSP with its own set of peripherals.
The devicetrees are organized to minimize duplication:
- Common perhiperals are described in sunxi-d1s-t113.dtsi
- DSP-related peripherals are described in sunxi-d1-t113.dtsi
- RISC-V specific hardware is described in sun20i-d1s.dtsi
- Functionality unique to the D1 variant is described in sun20i-d1.dtsi
The SOC_PERIPHERAL_IRQ macro handles the different #interrupt-cells
values between the ARM (GIC) and RISC-V (PLIC) versions of the SoC.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
Tested-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-5-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Several SoMs and boards are available that feature the Allwinner D1 or
D1s SoC. Document their compatible strings.
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Conor Dooley <conor.dooley@microchip.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Heiko Stuebner <heiko.stuebner@vrull.eu>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-4-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Some boards using the Allwinner D1 or D1s SoC are made by vendors not
previously documented.
Clockwork Tech LLC (https://www.clockworkpi.com/) manufactures the
ClockworkPi and DevTerm boards.
Beijing Widora Technology Co., Ltd. (https://mangopi.cc/) manufactures
the MangoPi family of boards.
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-3-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Allwinner sunxi SoCs with a RISC-V CPU use the sun20i designator. Match
that pattern in addition to the designators for 32 and 64-bit ARM SoCs.
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Heiko Stuebner <heiko@sntech.de>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Conor Dooley <conor.dooley@microchip.com>
Link: https://lore.kernel.org/r/20230126045738.47903-2-samuel@sholland.org
Signed-off-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Jakub Kicinski