Completely remove the old code for check_kfunc_call to help it work
with modules, and also the callback itself.
The previous commit adds infrastructure to register all sets and put
them in vmlinux or module BTF, and concatenates all related sets
organized by the hook and the type. Once populated, these sets remain
immutable for the lifetime of the struct btf.
Also, since we don't need the 'owner' module anywhere when doing
check_kfunc_call, drop the 'btf_modp' module parameter from
find_kfunc_desc_btf.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20220114163953.1455836-4-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
This patch prepares the kernel to support putting all kinds of kfunc BTF
ID sets in the struct btf itself. The various kernel subsystems will
make register_btf_kfunc_id_set call in the initcalls (for built-in code
and modules).
The 'hook' is one of the many program types, e.g. XDP and TC/SCHED_CLS,
STRUCT_OPS, and 'types' are check (allowed or not), acquire, release,
and ret_null (with PTR_TO_BTF_ID_OR_NULL return type).
A maximum of BTF_KFUNC_SET_MAX_CNT (32) kfunc BTF IDs are permitted in a
set of certain hook and type for vmlinux sets, since they are allocated
on demand, and otherwise set as NULL. Module sets can only be registered
once per hook and type, hence they are directly assigned.
A new btf_kfunc_id_set_contains function is exposed for use in verifier,
this new method is faster than the existing list searching method, and
is also automatic. It also lets other code not care whether the set is
unallocated or not.
Note that module code can only do single register_btf_kfunc_id_set call
per hook. This is why sorting is only done for in-kernel vmlinux sets,
because there might be multiple sets for the same hook and type that
must be concatenated, hence sorting them is required to ensure bsearch
in btf_id_set_contains continues to work correctly.
Next commit will update the kernel users to make use of this
infrastructure.
Finally, add __maybe_unused annotation for BTF ID macros for the
!CONFIG_DEBUG_INFO_BTF case, so that they don't produce warnings during
build time.
The previous patch is also needed to provide synchronization against
initialization for module BTF's kfunc_set_tab introduced here, as
described below:
The kfunc_set_tab pointer in struct btf is write-once (if we consider
the registration phase (comprised of multiple register_btf_kfunc_id_set
calls) as a single operation). In this sense, once it has been fully
prepared, it isn't modified, only used for lookup (from the verifier
context).
For btf_vmlinux, it is initialized fully during the do_initcalls phase,
which happens fairly early in the boot process, before any processes are
present. This also eliminates the possibility of bpf_check being called
at that point, thus relieving us of ensuring any synchronization between
the registration and lookup function (btf_kfunc_id_set_contains).
However, the case for module BTF is a bit tricky. The BTF is parsed,
prepared, and published from the MODULE_STATE_COMING notifier callback.
After this, the module initcalls are invoked, where our registration
function will be called to populate the kfunc_set_tab for module BTF.
At this point, BTF may be available to userspace while its corresponding
module is still intializing. A BTF fd can then be passed to verifier
using bpf syscall (e.g. for kfunc call insn).
Hence, there is a race window where verifier may concurrently try to
lookup the kfunc_set_tab. To prevent this race, we must ensure the
operations are serialized, or waiting for the __init functions to
complete.
In the earlier registration API, this race was alleviated as verifier
bpf_check_mod_kfunc_call didn't find the kfunc BTF ID until it was added
by the registration function (called usually at the end of module __init
function after all module resources have been initialized). If the
verifier made the check_kfunc_call before kfunc BTF ID was added to the
list, it would fail verification (saying call isn't allowed). The
access to list was protected using a mutex.
Now, it would still fail verification, but for a different reason
(returning ENXIO due to the failed btf_try_get_module call in
add_kfunc_call), because if the __init call is in progress the module
will be in the middle of MODULE_STATE_COMING -> MODULE_STATE_LIVE
transition, and the BTF_MODULE_LIVE flag for btf_module instance will
not be set, so the btf_try_get_module call will fail.
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20220114163953.1455836-3-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
While working on code to populate kfunc BTF ID sets for module BTF from
its initcall, I noticed that by the time the initcall is invoked, the
module BTF can already be seen by userspace (and the BPF verifier). The
existing btf_try_get_module calls try_module_get which only fails if
mod->state == MODULE_STATE_GOING, i.e. it can increment module reference
when module initcall is happening in parallel.
Currently, BTF parsing happens from MODULE_STATE_COMING notifier
callback. At this point, the module initcalls have not been invoked.
The notifier callback parses and prepares the module BTF, allocates an
ID, which publishes it to userspace, and then adds it to the btf_modules
list allowing the kernel to invoke btf_try_get_module for the BTF.
However, at this point, the module has not been fully initialized (i.e.
its initcalls have not finished). The code in module.c can still fail
and free the module, without caring for other users. However, nothing
stops btf_try_get_module from succeeding between the state transition
from MODULE_STATE_COMING to MODULE_STATE_LIVE.
This leads to a use-after-free issue when BPF program loads
successfully in the state transition, load_module's do_init_module call
fails and frees the module, and BPF program fd on close calls module_put
for the freed module. Future patch has test case to verify we don't
regress in this area in future.
There are multiple points after prepare_coming_module (in load_module)
where failure can occur and module loading can return error. We
illustrate and test for the race using the last point where it can
practically occur (in module __init function).
An illustration of the race:
CPU 0 CPU 1
load_module
notifier_call(MODULE_STATE_COMING)
btf_parse_module
btf_alloc_id // Published to userspace
list_add(&btf_mod->list, btf_modules)
mod->init(...)
... ^
bpf_check |
check_pseudo_btf_id |
btf_try_get_module |
returns true | ...
... | module __init in progress
return prog_fd | ...
... V
if (ret < 0)
free_module(mod)
...
close(prog_fd)
...
bpf_prog_free_deferred
module_put(used_btf.mod) // use-after-free
We fix this issue by setting a flag BTF_MODULE_F_LIVE, from the notifier
callback when MODULE_STATE_LIVE state is reached for the module, so that
we return NULL from btf_try_get_module for modules that are not fully
formed. Since try_module_get already checks that module is not in
MODULE_STATE_GOING state, and that is the only transition a live module
can make before being removed from btf_modules list, this is enough to
close the race and prevent the bug.
A later selftest patch crafts the race condition artifically to verify
that it has been fixed, and that verifier fails to load program (with
ENXIO).
Lastly, a couple of comments:
1. Even if this race didn't exist, it seems more appropriate to only
access resources (ksyms and kfuncs) of a fully formed module which
has been initialized completely.
2. This patch was born out of need for synchronization against module
initcall for the next patch, so it is needed for correctness even
without the aforementioned race condition. The BTF resources
initialized by module initcall are set up once and then only looked
up, so just waiting until the initcall has finished ensures correct
behavior.
Fixes: 541c3bad8d ("bpf: Support BPF ksym variables in kernel modules")
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20220114163953.1455836-2-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
I thought I was iterating over the array when actually the iteration is
over the values contained in the array?
Ugh, keep it simple.
Symptoms were a null deference in vfs_lock_file() when an NFSv3 client
that previously held a lock came back up and sent a notify.
Reported-by: Jonathan Woithe <jwoithe@just42.net>
Fixes: 7f024fcd5c ("Keep read and write fds with each nlm_file")
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
"Stolen memory" is memory set aside for use by an Intel integrated GPU.
The intel_graphics_quirks() early quirk reserves this memory when it is
called for a GPU that appears in the intel_early_ids[] table of integrated
GPUs.
Previously intel_graphics_quirks() was marked as QFLAG_APPLY_ONCE, so it
was called only for the first Intel GPU found. If a discrete GPU happened
to be enumerated first, intel_graphics_quirks() was called for it but not
for any integrated GPU found later. Therefore, stolen memory for such an
integrated GPU was never reserved.
For example, this problem occurs in this Alderlake-P (integrated) + DG2
(discrete) topology where the DG2 is found first, but stolen memory is
associated with the integrated GPU:
- 00:01.0 Bridge
`- 03:00.0 DG2 discrete GPU
- 00:02.0 Integrated GPU (with stolen memory)
Remove the QFLAG_APPLY_ONCE flag and call intel_graphics_quirks() for every
Intel GPU. Reserve stolen memory for the first GPU that appears in
intel_early_ids[].
[bhelgaas: commit log, add code comment, squash in
https://lore.kernel.org/r/20220118190558.2ququ4vdfjuahicm@ldmartin-desk2]
Link: https://lore.kernel.org/r/20220114002843.2083382-1-lucas.demarchi@intel.com
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org
Remove these unused ia32 compat declarations; all the bits involved have
either been withdrawn or hoisted to the VFS.
Signed-off-by: Darrick J. Wong <djwong@kernel.org>
Reviewed-by: Eric Sandeen <sandeen@redhat.com>
elf_mcount_loc and mcount_sort_thread definitions are not
initialized immediately within the function, which can cause
the judgment logic to use uninitialized values when the
initialization logic of subsequent code fails.
Link: https://lkml.kernel.org/r/20211212113358.34208-2-yinan@linux.alibaba.com
Link: https://lkml.kernel.org/r/20220118065241.42364-1-yinan@linux.alibaba.com
Fixes: 72b3942a17 ("scripts: ftrace - move the sort-processing in ftrace_init")
Tested-by: Nathan Chancellor <nathan@kernel.org>
Reviewed-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Yinan Liu <yinan@linux.alibaba.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When under stress, cleanup_net() can have to dismantle
netns in big numbers. ops_exit_list() currently calls
many helpers [1] that have no schedule point, and we can
end up with soft lockups, particularly on hosts
with many cpus.
Even for moderate amount of netns processed by cleanup_net()
this patch avoids latency spikes.
[1] Some of these helpers like fib_sync_up() and fib_sync_down_dev()
are very slow because net/ipv4/fib_semantics.c uses host-wide hash tables,
and ifindex is used as the only input of two hash functions.
ifindexes tend to be the same for all netns (lo.ifindex==1 per instance)
This will be fixed in a separate patch.
Fixes: 72ad937abd ("net: Add support for batching network namespace cleanups")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
bio_clone_fast() sets the cloned bio to have the same ->bi_bdev as the
source bio. This means that when request-based dm called setup_clone(),
the cloned bio had its ->bi_bdev pointing to the dm device. After Commit
0b6e522cdc ("blk-mq: use ->bi_bdev for I/O accounting")
__blk_account_io_start() started using the request's ->bio->bi_bdev for
I/O accounting, if it was set. This caused IO going to the underlying
devices to use the dm device for their I/O accounting.
Set up the proper ->bi_bdev in blk_rq_prep_clone based on the whole
device bdev for the queue the request is cloned onto.
Fixes: 0b6e522cdc ("blk-mq: use ->bi_bdev for I/O accounting")
Reported-by: Benjamin Marzinski <bmarzins@redhat.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
[hch: the commit message is mostly from a different patch from Benjamin]
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Reviewed-by: Benjamin Marzinski <bmarzins@redhat.com>
Link: https://lore.kernel.org/r/20220118070444.1241739-1-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Up to now cs35l41_hda_remove() returns zero unconditionally. Make it
return void instead which makes it easier to see in the callers that
there is no error to handle.
Also the return value of i2c and spi remove callbacks is ignored anyway.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220117220055.120955-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When possible use dev_err_probe help to properly deal with the
PROBE_DEFER error, the benefit is that DEFER issue will be logged
in the devices_deferred debugfs file.
Using dev_err_probe() can reduce code size, and the error value
gets printed.
Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220117160830.709403-5-tanureal@opensource.cirrus.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The test key now needs to be manually held when calling
cs35l41_register_errata_patch, after patch:
Add the missing function calls to this driver.
Fixes: f517ba4924 ("ASoC: cs35l41: Add support for hibernate memory retention mode")
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220117160830.709403-2-tanureal@opensource.cirrus.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
regmap_register_patch can't be used to apply the probe sequence as a
patch is already registers with the regmap by
cs35l41_register_errata_patch and only a single patch can be attached to
a single regmap. The driver doesn't currently rely on a cache sync to
re-apply this probe sequence so simply switch it to a multi write.
Fixes: 7b2f3eb492 ("ALSA: hda: cs35l41: Add support for CS35L41 in HDA systems")
Signed-off-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Lucas Tanure <tanureal@opensource.cirrus.com>
Link: https://lore.kernel.org/r/20220117160830.709403-1-tanureal@opensource.cirrus.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
When the target is a pid, not started by 'perf stat' we need to disable
the events, and in that case there is no need to setup affinities as we
use a dummy CPU map, with just one entry set to -1.
So stop doing it to avoid this needless call to sched_getaffinity():
# strace -ke sched_getaffinity perf stat -e cycles -p 241957 sleep 1
<SNIP>
sched_getaffinity(0, 512, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]) = 8
> /usr/lib64/libc-2.33.so(sched_getaffinity@@GLIBC_2.3.4+0x1a) [0xe6eea]
> /var/home/acme/bin/perf(affinity__setup+0x6a) [0x532a2a]
> /var/home/acme/bin/perf(__evlist__disable.constprop.0+0x27) [0x4b9827]
> /var/home/acme/bin/perf(cmd_stat+0x29b5) [0x431725]
> /var/home/acme/bin/perf(run_builtin+0x6a) [0x4a2cfa]
> /var/home/acme/bin/perf(main+0x612) [0x40f8c2]
> /usr/lib64/libc-2.33.so(__libc_start_main+0xd4) [0x27b74]
> /var/home/acme/bin/perf(_start+0x2d) [0x40fadd]
<SNIP>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20220117160931.1191712-5-acme@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
When the target is a pid, not started by 'perf stat' we need to enable
the events, and in that case there is no need to setup affinities as we
use a dummy CPU map, with just one entry set to -1.
So stop doing it to avoid this needless call to sched_getaffinity():
# strace -ke sched_getaffinity perf stat -e cycles -p 241957 sleep 1
<SNIP>
sched_getaffinity(0, 512, [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31]) = 8
> /usr/lib64/libc-2.33.so(sched_getaffinity@@GLIBC_2.3.4+0x1a) [0xe6eea]
> /var/home/acme/bin/perf(affinity__setup+0x6a) [0x5329ca]
> /var/home/acme/bin/perf(__evlist__enable.constprop.0+0x23) [0x4b9693]
> /var/home/acme/bin/perf(enable_counters+0x14d) [0x42de5d]
> /var/home/acme/bin/perf(cmd_stat+0x2358) [0x4310c8]
> /var/home/acme/bin/perf(run_builtin+0x6a) [0x4a2cfa]
> /var/home/acme/bin/perf(main+0x612) [0x40f8c2]
> /usr/lib64/libc-2.33.so(__libc_start_main+0xd4) [0x27b74]
> /var/home/acme/bin/perf(_start+0x2d) [0x40fadd]
<SNIP>
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20220117160931.1191712-4-acme@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
I.e. the simple:
$ perf stat sleep 1
Uses a dummy CPU map and thus there is no need to setup/cleanup
affinities to avoid IPIs, etc.
With this we're down to a sched_getaffinity() call, in the libnuma
initialization, that probably can be removed in a followup patch.
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20220117160931.1191712-3-acme@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Just like with free(), NULL is checked to avoid having all callers do
it.
Its convenient for when not using affinity setup/cleanup for dummy CPU
maps, i.e. CPU maps for pid targets.
Acked-by: Ian Rogers <irogers@google.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Ian Rogers <irogers@google.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: https://lore.kernel.org/r/20220117160931.1191712-2-acme@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Because of commit bf794bf52a ("powerpc/kprobes: Fix kallsyms
lookup across powerpc ABIv1 and ABIv2"), in ppc64 ABIv1, our perf
command eliminates the need to use the prefix "." at the symbol name.
But when the command "perf probe -a schedule" is executed on ppc64
ABIv1, it obtains two symbol address information through /proc/kallsyms,
for example:
cat /proc/kallsyms | grep -w schedule
c000000000657020 T .schedule
c000000000d4fdb8 D schedule
The symbol "D schedule" is not a function symbol, and perf will print:
"p:probe/schedule _text+13958584"Failed to write event: Invalid argument
Therefore, when searching symbols from map and adding probe point for
them, a symbol type check is added. If the type of symbol is not a
function, skip it.
Fixes: bf794bf52a ("powerpc/kprobes: Fix kallsyms lookup across powerpc ABIv1 and ABIv2")
Signed-off-by: Zechuan Chen <chenzechuan1@huawei.com>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jianlin Lv <Jianlin.Lv@arm.com>
Cc: Jin Yao <yao.jin@linux.intel.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Naveen N. Rao <naveen.n.rao@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Link: https://lore.kernel.org/r/20211228111338.218602-1-chenzechuan1@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Now that have_bytes is never modified, we can simplify this function.
First, we move the check for negative entropy_count to be first. That
ensures that subsequent reads of this will be non-negative. Then,
have_bytes and ibytes can be folded into their one use site in the
min_t() function.
Suggested-by: Dominik Brodowski <linux@dominikbrodowski.net>
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This is an old driver that has seen a lot of different eras of kernel
coding style. In an effort to make it easier to code for, unify the
coding style around the current norm, by accepting some of -- but
certainly not all of -- the suggestions from clang-format. This should
remove ambiguity in coding style, especially with regards to spacing,
when code is being changed or amended. Consequently it also makes code
review easier on the eyes, following one uniform style rather than
several.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This gets rid of another abstraction we no longer need. It would be nice
if we could instead make pool an array rather than a pointer, but the
latent entropy plugin won't be able to do its magic in that case. So
instead we put all accesses to the input pool's actual data through the
input_pool_data array directly.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The entropy estimator is calculated in terms of 1/8 bits, which means
there are various constants where things are shifted by 3. Move these
into our pool info enum with the other relevant constants. While we're
at it, move an English assertion about sizes into a proper BUILD_BUG_ON
so that the compiler can ensure this invariant.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
The other pool constants are prepended with POOL_, but not these last
ones. Rename them. This will then let us move them into the enum in the
following commit.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
We already had the POOL_* constants, so deduplicate the older INPUT_POOL
ones. As well, fold EXTRACT_SIZE into the poolinfo enum, since it's
related.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
We no longer have an output pool. Rather, we have just a wakeup bits
threshold for /dev/random reads, presumably so that processes don't
hang. This value, random_write_wakeup_bits, is configurable anyway. So
all the no longer usefully named OUTPUT_POOL constants were doing was
setting a reasonable default for random_write_wakeup_bits. This commit
gets rid of the constants and just puts it all in the default value of
random_write_wakeup_bits.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Originally, the RNG used several pools, so having things abstracted out
over a generic entropy_store object made sense. These days, there's only
one input pool, and then an uneven mix of usage via the abstraction and
usage via &input_pool. Rather than this uneasy mixture, just get rid of
the abstraction entirely and have things always use the global. This
simplifies the code and makes reading it a bit easier.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
This argument is always set to zero, as a result of us not caring about
keeping a certain amount reserved in the pool these days. So just remove
it and cleanup the function signatures.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
There were a few things added under the "if (fips_enabled)" banner,
which never really got completed, and the FIPS people anyway are
choosing a different direction. Rather than keep around this halfbaked
code, get rid of it so that we can focus on a single design of the RNG
rather than two designs.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Rather than using the userspace type, __uXX, switch to using uXX. And
rather than using variously chosen `char *` or `unsigned char *`, use
`u8 *` uniformly for things that aren't strings, in the case where we
are doing byte-by-byte traversal.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Now that we're only using one polynomial, we can cleanup its
representation into constants, instead of passing around pointers
dynamically to select different polynomials. This improves the codegen
and makes the code a bit more straightforward.
Reviewed-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
With SHA-1 no longer being used for anything performance oriented, and
also soon to be phased out entirely, we can make up for the space added
by unrolled BLAKE2s by simply re-rolling SHA-1. Since SHA-1 is so much
more complex, re-rolling it more or less takes care of the code size
added by BLAKE2s. And eventually, hopefully we'll see SHA-1 removed
entirely from most small kernel builds.
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Ard Biesheuvel <ardb@kernel.org>
Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Basically nobody should use blake2s in an HMAC construction; it already
has a keyed variant. But unfortunately for historical reasons, Noise,
used by WireGuard, uses HKDF quite strictly, which means we have to use
this. Because this really shouldn't be used by others, this commit moves
it into wireguard's noise.c locally, so that kernels that aren't using
WireGuard don't get this superfluous code baked in. On m68k systems,
this shaves off ~314 bytes.
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Tested-by: Geert Uytterhoeven <geert@linux-m68k.org>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Commit 6048fdcc5f ("lib/crypto: blake2s: include as built-in") took
away a number of prompt texts from other crypto libraries. This makes
values flip from built-in to module when oldconfig runs, and causes
problems when these crypto libs need to be built in for thingslike
BIG_KEYS.
Fixes: 6048fdcc5f ("lib/crypto: blake2s: include as built-in")
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: linux-crypto@vger.kernel.org
Signed-off-by: Justin M. Forbes <jforbes@fedoraproject.org>
[Jason: - moved menu into submenu of lib/ instead of root menu
- fixed chacha sub-dependencies for CONFIG_CRYPTO]
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
New support:
- DMA_MEMCPY_SG support is bought back as we have a user in Xilinx driver
- Support for TI J721S2 SoC in k3-udma driver
- Support for Ingenic MDMA and BDMA in the JZ4760
- Support for Renesas r8a779f0 dmac
Updates:
- We are finally getting rid of slave_id, so this brings in the changes
across tree for that using tag dmaengine_topic_slave_id_removal_5.17
- updates for idxd driver
- at_xdmac driver cleanup
-----BEGIN PGP SIGNATURE-----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=1KYH
-----END PGP SIGNATURE-----
Merge tag 'dmaengine-5.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine
Pull dmaengine updates from Vinod Koul:
"A bunch of new support and few updates to drivers:
New support:
- DMA_MEMCPY_SG support is bought back as we have a user in Xilinx
driver
- Support for TI J721S2 SoC in k3-udma driver
- Support for Ingenic MDMA and BDMA in the JZ4760
- Support for Renesas r8a779f0 dmac
Updates:
- We are finally getting rid of slave_id, so this brings in the
changes across tree for that
- updates for idxd driver
- at_xdmac driver cleanup"
* tag 'dmaengine-5.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vkoul/dmaengine: (60 commits)
dt-bindings: dma-controller: Split interrupt fields in example
dmaengine: pch_dma: Remove usage of the deprecated "pci-dma-compat.h" API
dmaengine: at_xdmac: Fix race over irq_status
dmaengine: at_xdmac: Remove a level of indentation in at_xdmac_tasklet()
dmaengine: at_xdmac: Fix at_xdmac_lld struct definition
dmaengine: at_xdmac: Fix lld view setting
dmaengine: at_xdmac: Remove a level of indentation in at_xdmac_advance_work()
dmaengine: at_xdmac: Fix concurrency over xfers_list
dmaengine: at_xdmac: Move the free desc to the tail of the desc list
dmaengine: at_xdmac: Fix race for the tx desc callback
dmaengine: at_xdmac: Fix concurrency over chan's completed_cookie
dmaengine: at_xdmac: Print debug message after realeasing the lock
dmaengine: at_xdmac: Start transfer for cyclic channels in issue_pending
dmaengine: at_xdmac: Don't start transactions at tx_submit level
dmaengine: idxd: deprecate token sysfs attributes for read buffers
dmaengine: idxd: change bandwidth token to read buffers
dmaengine: idxd: fix wq settings post wq disable
dmaengine: idxd: change MSIX allocation based on per wq activation
dmaengine: idxd: fix descriptor flushing locking
dmaengine: idxd: embed irq_entry in idxd_wq struct
...
since commit 2279f540ea ("sched/deadline: Fix priority
inheritance with multiple scheduling classes"), we should not
keep it here.
Signed-off-by: Hui Su <suhui_kernel@163.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Daniel Bristot de Oliveira <bristot@redhat.com>
Link: https://lore.kernel.org/r/20220107095254.GA49258@localhost.localdomain
For PREEMPT/DYNAMIC_PREEMPT the *_unlock() will already trigger a
preemption, no point in then calling preempt_schedule_common()
*again*.
Use _cond_resched() instead, since this is a NOP for the preemptible
configs while it provide a preemption point for the others.
Reported-by: xuhaifeng <xuhaifeng@oppo.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/YcGnvDEYBwOiV0cR@hirez.programming.kicks-ass.net
Quieten all kernel-doc warnings in kernel/sched/fair.c:
kernel/sched/fair.c:3663: warning: No description found for return value of 'update_cfs_rq_load_avg'
kernel/sched/fair.c:8601: warning: No description found for return value of 'asym_smt_can_pull_tasks'
kernel/sched/fair.c:8673: warning: Function parameter or member 'sds' not described in 'update_sg_lb_stats'
kernel/sched/fair.c:9483: warning: contents before sections
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com>
Acked-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20211218055900.2704-1-rdunlap@infradead.org
There are two types of forced idle time: forced idle time from cookie'd
task and forced idle time form uncookie'd task. The forced idle time from
uncookie'd task is actually caused by the cookie'd task in runqueue
indirectly, and it's more accurate to measure the capacity loss with the
sum of both.
Assuming cpu x and cpu y are a pair of SMT siblings, consider the
following scenarios:
1.There's a cookie'd task running on cpu x, and there're 4 uncookie'd
tasks running on cpu y. For cpu x, there will be 80% forced idle time
(from uncookie'd task); for cpu y, there will be 20% forced idle time
(from cookie'd task).
2.There's a uncookie'd task running on cpu x, and there're 4 cookie'd
tasks running on cpu y. For cpu x, there will be 80% forced idle time
(from cookie'd task); for cpu y, there will be 20% forced idle time
(from uncookie'd task).
The scenario1 can recurrent by stress-ng(scenario2 can recurrent similary):
(cookie'd)taskset -c x stress-ng -c 1 -l 100
(uncookie'd)taskset -c y stress-ng -c 4 -l 100
In the above two scenarios, the total capacity loss is 1 cpu, but in
scenario1, the cookie'd forced idle time tells us 20% cpu capacity loss, in
scenario2, the cookie'd forced idle time tells us 80% cpu capacity loss,
which are not accurate. It'll be more accurate to measure with cookie'd
forced idle time and uncookie'd forced idle time.
Signed-off-by: Cruz Zhao <CruzZhao@linux.alibaba.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Josh Don <joshdon@google.com>
Link: https://lore.kernel.org/r/1641894961-9241-2-git-send-email-CruzZhao@linux.alibaba.com
Similarly to util_avg and util_sum, don't sync load_sum with the low
bound of load_avg but only ensure that load_sum stays in the correct range.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Tested-by: Sachin Sant <sachinp@linux.ibm.com>
Link: https://lkml.kernel.org/r/20220111134659.24961-5-vincent.guittot@linaro.org
Similarly to util_avg and util_sum, don't sync runnable_sum with the low
bound of runnable_avg but only ensure that runnable_sum stays in the
correct range.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Tested-by: Sachin Sant <sachinp@linux.ibm.com>
Link: https://lkml.kernel.org/r/20220111134659.24961-4-vincent.guittot@linaro.org
Rick reported performance regressions in bugzilla because of cpu frequency
being lower than before:
https://bugzilla.kernel.org/show_bug.cgi?id=215045
He bisected the problem to:
commit 1c35b07e6d ("sched/fair: Ensure _sum and _avg values stay consistent")
This commit forces util_sum to be synced with the new util_avg after
removing the contribution of a task and before the next periodic sync. By
doing so util_sum is rounded to its lower bound and might lost up to
LOAD_AVG_MAX-1 of accumulated contribution which has not yet been
reflected in util_avg.
update_tg_cfs_util() is not the only place where we round util_sum and
lost some accumulated contributions that are not already reflected in
util_avg. Modify update_tg_cfs_util() and detach_entity_load_avg() to not
sync util_sum with the new util_avg. Instead of always setting util_sum to
the low bound of util_avg, which can significantly lower the utilization,
we propagate the difference. In addition, we also check that cfs's util_sum
always stays above the lower bound for a given util_avg as it has been
observed that sched_entity's util_sum is sometimes above cfs one.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Tested-by: Sachin Sant <sachinp@linux.ibm.com>
Link: https://lkml.kernel.org/r/20220111134659.24961-3-vincent.guittot@linaro.org
Rick reported performance regressions in bugzilla because of cpu frequency
being lower than before:
https://bugzilla.kernel.org/show_bug.cgi?id=215045
He bisected the problem to:
commit 1c35b07e6d ("sched/fair: Ensure _sum and _avg values stay consistent")
This commit forces util_sum to be synced with the new util_avg after
removing the contribution of a task and before the next periodic sync. By
doing so util_sum is rounded to its lower bound and might lost up to
LOAD_AVG_MAX-1 of accumulated contribution which has not yet been
reflected in util_avg.
Instead of always setting util_sum to the low bound of util_avg, which can
significantly lower the utilization of root cfs_rq after propagating the
change down into the hierarchy, we revert the change of util_sum and
propagate the difference.
In addition, we also check that cfs's util_sum always stays above the
lower bound for a given util_avg as it has been observed that
sched_entity's util_sum is sometimes above cfs one.
Fixes: 1c35b07e6d ("sched/fair: Ensure _sum and _avg values stay consistent")
Reported-by: Rick Yiu <rickyiu@google.com>
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Tested-by: Sachin Sant <sachinp@linux.ibm.com>
Link: https://lkml.kernel.org/r/20220111134659.24961-2-vincent.guittot@linaro.org
With write operation on psi files replacing old trigger with a new one,
the lifetime of its waitqueue is totally arbitrary. Overwriting an
existing trigger causes its waitqueue to be freed and pending poll()
will stumble on trigger->event_wait which was destroyed.
Fix this by disallowing to redefine an existing psi trigger. If a write
operation is used on a file descriptor with an already existing psi
trigger, the operation will fail with EBUSY error.
Also bypass a check for psi_disabled in the psi_trigger_destroy as the
flag can be flipped after the trigger is created, leading to a memory
leak.
Fixes: 0e94682b73 ("psi: introduce psi monitor")
Reported-by: syzbot+cdb5dd11c97cc532efad@syzkaller.appspotmail.com
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Analyzed-by: Eric Biggers <ebiggers@kernel.org>
Signed-off-by: Suren Baghdasaryan <surenb@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20220111232309.1786347-1-surenb@google.com
Some hypervisors support Arch LBR, but without the LBR XSAVE support.
The current Arch LBR init code prints a warning when the xsave size (0) is
unexpected. Avoid printing the warning for the "no LBR XSAVE" case.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20211215204029.150686-1-ak@linux.intel.com
Current ADL uncore code only supports the legacy IMC (memory controller)
free-running counters. Besides the free-running counters, ADL also
supports several general purpose-counters.
The general-purpose counters can also be accessed via MMIO but in a
different location. Factor out __uncore_imc_init_box() with offset as a
parameter. The function can be shared between ADL and TGL.
The event format and the layout of the control registers are a little
bit different from other uncore counters.
The intel_generic_uncore_mmio_enable_event() can be shared with client
IMC uncore. Expose the function.
Add more PCI IDs for ADL machines.
Fixes: 772ed05f3c ("perf/x86/intel/uncore: Add Alder Lake support")
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/1642111554-118524-1-git-send-email-kan.liang@linux.intel.com
Using static_branch to replace the LBR INFO flags to optimize the LBR
INFO parsing.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Kan Liang <kan.liang@linux.intel.com>
Link: https://lkml.kernel.org/r/1641315077-96661-2-git-send-email-peterz@infradead.org
