Samsung Exynos SoC reuses several devices from older designs, thus
historically we kept the old (block's) compatible only. This works fine
and there is no bug here, however guidelines expressed in
Documentation/devicetree/bindings/writing-bindings.rst state that:
1. Compatibles should be specific.
2. We should add new compatibles in case of bugs or features.
Add compatibles specific to each SoC in front of all old-SoC-like
compatibles.
Reviewed-by: Alim Akhtar <alim.akhtar@samsung.com>
Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20231108104343.24192-6-krzysztof.kozlowski@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Cited commit removed the strscpy() call and kept the snprintf() only.
It is common to use 'dev->name' as the format string before a netdev is
registered, this results in 'res' and 'name' pointers being equal.
According to POSIX, if copying takes place between objects that overlap
as a result of a call to sprintf() or snprintf(), the results are
undefined.
Add back the strscpy() and use 'buf' as an intermediate buffer.
Fixes: 7ad17b04dc ("net: trust the bitmap in __dev_alloc_name()")
Cc: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Vlad Buslov <vladbu@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When in the list_for_each_entry iteration, reload of p->state->settings
with a local setting from old_state will turn the list iteration into an
infinite loop.
The typical symptom when the issue happens, will be a printk message like:
"not freeing pin xx (xxx) as part of deactivating group xxx - it is
already used for some other setting".
This is a compiler-dependent problem, one instance occurred using Clang
version 10.0 on the arm64 architecture with linux version 4.19.
Fixes: 6e5e959dde ("pinctrl: API changes to support multiple states per device")
Signed-off-by: Maria Yu <quic_aiquny@quicinc.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20231115102824.23727-1-quic_aiquny@quicinc.com
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
The top syzbot report for networking (#14 for the entire kernel)
is the queue timeout splat. We kept it around for a long time,
because in real life it provides pretty strong signal that
something is wrong with the driver or the device.
Removing it is also likely to break monitoring for those who
track it as a kernel warning.
Nevertheless, WARN()ings are best suited for catching kernel
programming bugs. If a Tx queue gets starved due to a pause
storm, priority configuration, or other weirdness - that's
obviously a problem, but not a problem we can fix at
the kernel level.
Bite the bullet and convert the WARN() to a print.
Before:
NETDEV WATCHDOG: eni1np1 (netdevsim): transmit queue 0 timed out 1975 ms
WARNING: CPU: 0 PID: 0 at net/sched/sch_generic.c:525 dev_watchdog+0x39e/0x3b0
[... completely pointless stack trace of a timer follows ...]
Now:
netdevsim netdevsim1 eni1np1: NETDEV WATCHDOG: CPU: 0: transmit queue 0 timed out 1769 ms
Alternatively we could mark the drivers which syzbot has
learned to abuse as "print-instead-of-WARN" selectively.
Reported-by: syzbot+d55372214aff0faa1f1f@syzkaller.appspotmail.com
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan says:
====================
bnxt_en: TX path improvements
All patches in this patchset are related to improving the TX path.
There are 2 areas of improvements:
1. The TX interrupt logic currently counts the number of TX completions
to determine the number of TX SKBs to free. We now change it so that
the TX completion will now contain the hardware consumer index
information. The driver will keep track of the latest hardware
consumer index from the last TX completion and clean up all TX SKBs
up to that index. This scheme aligns better with future chips and
allows xmit_more code path to be more optimized.
2. The current driver logic requires an additional MSIX for each
additional MQPRIO TX ring. This scheme uses too many MSIX vectors if
the user enables a large number of MQPRIO TCs. We now use a new scheme
that will use the same MSIX for all the MQPRIO TX rings for each
ethtool channel. Each ethtool TX channel can have up to 8 MQPRIO
TX rings and now they all will share the same MSIX.
v2: Rebased
v1 posted on Oct 27 2023 right before the close of net-next:
https://lore.kernel.org/netdev/20231027232252.36111-1-michael.chan@broadcom.com/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that we use the cumulative consumer index scheme for TX completion,
we don't need to have one TX completion per TX packet in the xmit_more
code path. Set the TX_BD_FLAGS_NO_CMPL flag if xmit_more is true.
Fallback to one interrupt per packet if the ring is filled beyond
bp->tx_wake_thresh.
Also, move the wmb() to bnxt_txr_db_kick(). When xmit_more is true,
we'll skip the bnxt_txr_db_kick() call and there is no need to call
wmb() to sync. the TX BD data.
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We can now fully support sharing the same MSIX for all mqprio TX rings
belonging to the same ethtool channel with the new infrastructure:
1. Allocate the proper entries for cp_ring_arr in struct bnxt_cp_ring_info
to support the additional TX rings.
2. Populate the tx_ring array in struct bnxt_napi for all TX rings
sharing the same NAPI.
3. bnxt_num_tx_to_cp() returns the proper NQ/completion rings to support
the TX rings in the input.
4. Adjust bnxt_get_num_ring_stats() for the reduced number of ring
counters with the new scheme.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add 3 macros that handle to conversions between TC numbers and TX
ring numbers. These will help to clarify the existing logic and the
new logic in the next patch.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Up until now, each TX ring always requires a completion ring/NQ/MSIX.
bnxt_trim_rings() and the assignment of bp->cp_nr_rings always make
this assumption. This will no longer be true in the next patches, so
we refactor and add helper functions to determine the proper relationship
between TX rings and the required completion ring/NQ/MSIX. This patch
does not change the 1:1 relationship yet.
Note that on P5 chips, each RX and TX ring still requires a completion
ring. Only the number of NQs has been reduced. We should no longer call
bnxt_trim_rings() to adjust the RX and TX rings on P5 chips. Replace with
simple logic to check that RX + TX < CP and adjust accordingly.
bnxt_check_rings() should call _bnxt_get_max_rings() to get the raw
number of rings instead of bnxt_get_max_rings(). If we are about to
create TCs, bnxt_get_max_rings() would not be able to calculate the max
rings correctly.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For each mqprio TC, we allocate a set of TX rings to map to the new
hardware CoS queue. Expand the tx_ring pointer in struct bnxt_napi
to an array of 8 to support up to 8 TX rings, one for each TC.
Only array entry 0 is used at this time. The rest of the array
entries will be used in later patches.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add 2 helper functions to set coalescing for each RX and TX rings. This
will make it easier to expand the number of TX rings per MSIX in the
next patches.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to support multiple TX rings on the same MSIX, we'll use the
upper byte of the TX opaque field to store the ring index in the new
tx_napi_idx field. This tx_napi_idx field is currently always 0 until
more infrastructure is added in later patches.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
bnxt_tx_int() processes the only one TX ring from the bnxt_napi pointer.
To prepare for more TX rings associated with the bnxt_napi structure,
add a new __bnxt_tx_int() function that takes the bnxt_tx_ring_info
pointer to process that one TX ring. No functional change.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These 2 constants were used for the one RX and one TX completion
ring pointer in the cpr->cp_ring_arr fixed array. Now that we've
changed to allocating the array for the exact number of entries to
support more TX rings, we no longer use these constants.
The array index as well as the type of completion ring (RX/TX) are
now encoded in the handle for the completion ring. This will allow
us to locate the completion ring during NAPI for any number of
completion rings sharing the same MSIX. In the following patches,
we'll be adding support for more TX rings associated with the same
MSIX vector.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
From the TX or RX ring structure, we need to find the corresponding
completion ring during initialization. On P5 chips, we use the MSIX/napi
entry to locate the completion ring because there is only one RX/TX
ring per MSIX. To allow multiple TX rings for each MSIX, we need
to add a direct pointer from the TX ring and RX ring structures.
This also simplifies the existing logic.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The cp_ring_arr is currently a fixed array of 2 pointers for the
TX and RX completion rings. These pointers are allocated during
ring initialization. Currntly, we support up to 2 completion rings
for each MSIX. In order to support more completion rings, we change
this fixed array to a pointer and allocate the required entries
during ring initialization. This patch keeps the current scheme of
allocating only 2 entries when needed. Later patches will expand
and allocate more entries when required.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
From the TX or RX ring structure, we need to find the corresponding
completion ring during initialization. On P5 chips, we use the MSIX/napi
entry to locate the completion ring because there is only one RX/TX
ring per MSIX. To allow multiple TX rings for each MSIX, we need
to add a direct pointer from the TX ring and RX ring structures.
This also simplifies the existing logic.
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, the opaque field in the TX BD is only used for debugging.
The TX completion logic relies on getting one TX completion for each
packet and they always complete in order.
Improve this scheme by putting the producer information (ring index plus
number of BDs for the packet) in the opaque field. This way, we can
handle TX completion processing by looking at the last TX completion
instead of counting the number of completions.
Since we no longer need to count the exact number of completions, we can
optimize xmit_more by disabling TX completion when the xmit_more
condition is true. This will be done in later patches.
This patch is only initializing the opaque field in the TX BD and is
not changing the driver's TX completion logic yet.
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The usage count of struct dev_pm_info is an implementation detail that
is only available if CONFIG_PM is enabled, so printing it in a debug message
causes a build failure in configurations without PM:
In file included from include/linux/device.h:15,
from include/linux/pci.h:37,
from drivers/accel/ivpu/ivpu_pm.c:8:
drivers/accel/ivpu/ivpu_pm.c: In function 'ivpu_rpm_get_if_active':
drivers/accel/ivpu/ivpu_pm.c:254:51: error: 'struct dev_pm_info' has no member named 'usage_count'
254 | atomic_read(&vdev->drm.dev->power.usage_count));
| ^
include/linux/dev_printk.h:129:48: note: in definition of macro 'dev_printk'
129 | _dev_printk(level, dev, fmt, ##__VA_ARGS__); \
| ^~~~~~~~~~~
drivers/accel/ivpu/ivpu_drv.h:75:17: note: in expansion of macro 'dev_dbg'
75 | dev_dbg((vdev)->drm.dev, "[%s] " fmt, #type, ##args); \
| ^~~~~~~
drivers/accel/ivpu/ivpu_pm.c:253:9: note: in expansion of macro 'ivpu_dbg'
253 | ivpu_dbg(vdev, RPM, "rpm_get_if_active count %d\n",
| ^~~~~~~~
The print message does not seem essential, so the easiest workaround is
to just remove it.
Fixes: c39dc15191 ("accel/ivpu: Read clock rate only if device is up")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Signed-off-by: Stanislaw Gruszka <stanislaw.gruszka@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231027152633.528490-1-arnd@kernel.org
(cherry picked from commit 1470acbef1)
Signed-off-by: Jacek Lawrynowicz <jacek.lawrynowicz@linux.intel.com>
When moving the *-internal-delay-ps properties to only apply for RGMII
interface modes there where a typo in the text formatting.
Signed-off-by: Niklas Söderlund <niklas.soderlund+renesas@ragnatech.se>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit 3780bb2931.
The cited commit introduced unwanted behavior.
The intent for the commit was to be able to detect carrier loss/gain
for just the NIC connected to the BMC. The unwanted effect is a
carrier loss for auxiliary paths also causes the BMC to lose
carrier. The BMC never regains carrier despite the secondary NIC
regaining a link.
This change, when merged, needs to be backported to stable kernels.
5.4-stable, 5.10-stable, 5.15-stable, 6.1-stable, 6.5-stable
Fixes: 3780bb2931 ("ncsi: Propagate carrier gain/loss events to the NCSI controller")
CC: stable@vger.kernel.org
Signed-off-by: Johnathan Mantey <johnathanx.mantey@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adds:
- DEFINE_GUARD_COND() / DEFINE_LOCK_GUARD_1_COND() to extend existing
guards with conditional lock primitives, eg. mutex_trylock(),
mutex_lock_interruptible().
nb. both primitives allow NULL 'locks', which cause the lock to
fail (obviously).
- extends scoped_guard() to not take the body when the the
conditional guard 'fails'. eg.
scoped_guard (mutex_intr, &task->signal_cred_guard_mutex) {
...
}
will only execute the body when the mutex is held.
- provides scoped_cond_guard(name, fail, args...); which extends
scoped_guard() to do fail when the lock-acquire fails.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20231102110706.460851167%40infradead.org
Trying to avoid that didn't bring much value after testing, add comment
about this.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Rafael J. Wysocki <rafael@kernel.org>
Link: https://lkml.kernel.org/r/20231114193840.4041-3-frederic@kernel.org
Add missing explanation concerning IRQs re-enablement constraints in
the cpuidle path against timers.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Rafael J. Wysocki <rafael@kernel.org>
Link: https://lkml.kernel.org/r/20231114193840.4041-2-frederic@kernel.org
Low priority tasks (e.g., SCHED_OTHER) can suffer starvation if tasks
with higher priority (e.g., SCHED_FIFO) monopolize CPU(s).
RT Throttling has been introduced a while ago as a (mostly debug)
countermeasure one can utilize to reserve some CPU time for low priority
tasks (usually background type of work, e.g. workqueues, timers, etc.).
It however has its own problems (see documentation) and the undesired
effect of unconditionally throttling FIFO tasks even when no lower
priority activity needs to run (there are mechanisms to fix this issue
as well, but, again, with their own problems).
Introduce deadline servers to service low priority tasks needs under
starvation conditions. Deadline servers are built extending SCHED_DEADLINE
implementation to allow 2-level scheduling (a sched_deadline entity
becomes a container for lower priority scheduling entities).
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/4968601859d920335cf85822eb573a5f179f04b8.1699095159.git.bristot@kernel.org
In preparation of introducing !task sched_dl_entity; move the
bandwidth accounting into {en.de}queue_dl_entity().
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Phil Auld <pauld@redhat.com>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Link: https://lkml.kernel.org/r/a86dccbbe44e021b8771627e1dae01a69b73466d.1699095159.git.bristot@kernel.org
Create a single function that initializes a sched_dl_entity.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Phil Auld <pauld@redhat.com>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Link: https://lkml.kernel.org/r/51acc695eecf0a1a2f78f9a044e11ffd9b316bcf.1699095159.git.bristot@kernel.org
Now that trace_sched_stat_runtime() no longer takes a vruntime
argument, the task specific bits are identical between
update_curr_common() and update_curr().
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tracing the runtime delta makes sense, observer can sum over time.
Tracing the absolute vruntime makes less sense, inconsistent:
absolute-vs-delta, but also vruntime delta can be computed from
runtime delta.
Removing the vruntime thing also makes the two tracepoint sites
identical, allowing to unify the code in a later patch.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
All classes use sched_entity::exec_start to track runtime and have
copies of the exact same code around to compute runtime.
Collapse all that.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Phil Auld <pauld@redhat.com>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Reviewed-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/54d148a144f26d9559698c4dd82d8859038a7380.1699095159.git.bristot@kernel.org
Since the RB-tree is now sorted by deadline, let's first try the
leftmost entity which has the earliest virtual deadline. I've done
some benchmarks to see its effectiveness.
All the benchmarks are done inside a normal cpu cgroup in a clean
environment with cpu turbo disabled, on a dual-CPU Intel Xeon(R)
Platinum 8260 with 2 NUMA nodes each of which has 24C/48T.
hackbench: process/thread + pipe/socket + 1/2/4/8 groups
netperf: TCP/UDP + STREAM/RR + 24/48/72/96/192 threads
tbench: loopback 24/48/72/96/192 threads
schbench: 1/2/4/8 mthreads
direct: cfs_rq has only one entity
parity: RUN_TO_PARITY
fast: O(1) fastpath
slow: heap search
(%) direct parity fast slow
hackbench 92.95 2.02 4.91 0.12
netperf 68.08 6.60 24.18 1.14
tbench 67.55 11.22 20.61 0.62
schbench 69.91 2.65 25.73 1.71
The above results indicate that this fastpath really makes task
selection more efficient.
Signed-off-by: Abel Wu <wuyun.abel@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20231115033647.80785-4-wuyun.abel@bytedance.com
Sort the task timeline by virtual deadline and keep the min_vruntime
in the augmented tree, so we can avoid doubling the worst case cost
and make full use of the cached leftmost node to enable O(1) fastpath
picking in next patch.
Signed-off-by: Abel Wu <wuyun.abel@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20231115033647.80785-3-wuyun.abel@bytedance.com
Since commit fc137c0dda ("sched/numa: enhance vma scanning logic")
NUMA Balancing allows updating PTEs to trap NUMA hinting faults if the
task had previously accessed VMA. However unconditional scan of VMAs are
allowed during initial phase of VMA creation until process's
mm numa_scan_seq reaches 2 even though current task had not accessed VMA.
Rationale:
- Without initial scan subsequent PTE update may never happen.
- Give fair opportunity to all the VMAs to be scanned and subsequently
understand the access pattern of all the VMAs.
But it has a corner case where, if a VMA is created after some time,
process's mm numa_scan_seq could be already greater than 2.
For e.g., values of mm numa_scan_seq when VMAs are created by running
mmtest autonuma benchmark briefly looks like:
start_seq=0 : 459
start_seq=2 : 138
start_seq=3 : 144
start_seq=4 : 8
start_seq=8 : 1
start_seq=9 : 1
This results in no unconditional PTE updates for those VMAs created after
some time.
Fix:
- Note down the initial value of mm numa_scan_seq in per VMA start_seq.
- Allow unconditional scan till start_seq + 2.
Result:
SUT: AMD EPYC Milan with 2 NUMA nodes 256 cpus.
base kernel: upstream 6.6-rc6 with Mels patches [1] applied.
kernbench
========== base patched %gain
Amean elsp-128 165.09 ( 0.00%) 164.78 * 0.19%*
Duration User 41404.28 41375.08
Duration System 9862.22 9768.48
Duration Elapsed 519.87 518.72
Ops NUMA PTE updates 1041416.00 831536.00
Ops NUMA hint faults 263296.00 220966.00
Ops NUMA pages migrated 258021.00 212769.00
Ops AutoNUMA cost 1328.67 1114.69
autonumabench
NUMA01_THREADLOCAL
==================
Amean elsp-NUMA01_THREADLOCAL 81.79 (0.00%) 67.74 * 17.18%*
Duration User 54832.73 47379.67
Duration System 75.00 185.75
Duration Elapsed 576.72 476.09
Ops NUMA PTE updates 394429.00 11121044.00
Ops NUMA hint faults 1001.00 8906404.00
Ops NUMA pages migrated 288.00 2998694.00
Ops AutoNUMA cost 7.77 44666.84
Signed-off-by: Raghavendra K T <raghavendra.kt@amd.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Link: https://lore.kernel.org/r/2ea7cbce80ac7c62e90cbfb9653a7972f902439f.1697816692.git.raghavendra.kt@amd.com
Instead of the IRQ number user the struct irq_info pointer as parameter
in the internal pirq related functions. This allows to drop some calls
of info_for_irq().
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
Modify the internal bind- and unbind-interfaces to take a struct
irq_info parameter. When allocating a new IRQ pass the pointer from
the allocating function further up.
This will reduce the number of info_for_irq() calls and make the code
more efficient.
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
Instead of having a common function for allocating a single IRQ or a
consecutive number of IRQs, split up the functionality into the callers
of xen_allocate_irqs_dynamic().
This allows to handle any allocation error in xen_irq_init() gracefully
instead of panicing the system. Let xen_irq_init() return the irq_info
pointer or NULL in case of an allocation error.
Additionally set the IRQ into irq_info already at allocation time, as
otherwise the IRQ would be '0' (which is a valid IRQ number) until
being set.
Signed-off-by: Juergen Gross <jgross@suse.com>
Reviewed-by: Oleksandr Tyshchenko <oleksandr_tyshchenko@epam.com>
Signed-off-by: Juergen Gross <jgross@suse.com>
FIFO underruns are observed when FBC is enabled on plane 2 or
plane 3. Recommended WA is to update the FBC enabling sequence.
The plane binding register bits need to be updated separately
before programming the FBC enable bit.
Bspec: 74151
Reviewed-by: Mika Kahola <mika.kahola@intel.com> #v3
Signed-off-by: Vinod Govindapillai <vinod.govindapillai@intel.com>
Reviewed-by: Mika Kahola <mika.kahola@intel.com>
Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20231111114320.87277-2-vinod.govindapillai@intel.com
In percpu reader mode, trylock() for read had a lost wakeup: on failure
to get the lock, we may have caused a writer to fail to get the lock,
because we temporarily elevated the reader count.
We need to check for waiters after decrementing the read count - not
before.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
In no_data_io mode, we expect data checksums to be wrong - don't want to
spew the log with them.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
When searching the link table for the matching inode, we were searching
for a specific - incorrect - snapshot ID as well, causing us to fail to
find the inode.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
Running with fewer max btree paths doesn't work anymore when replication
is enabled - as we've added e.g. the freespace and bucket gens btrees,
we naturally end up needing more btree paths.
This is an issue with lockdep, we end up taking more locks than lockdep
will track (the MAX_LOCKD_DEPTH constant). But bcachefs as merged does
not yet support lockdep anyways, so we can leave that for later.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
The journal read path had some informational log statements preperatory
for ZNS support - they're not of interest to users, so we can turn them
off.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
In may_delete_deleted_inode(), there's a corner case when a snapshot was
taken while we had an unlinked inode: we don't want to delete the inode
in the internal (shared) snapshot node, since it might have been
reattached in a descendent snapshot.
Instead we propagate the key to any snapshot leaves it doesn't exist in,
so that it can be deleted there if necessary, and then clear the
unlinked flag in the internal node.
But we forgot to commit after clearing the unlinked flag, causing us to
go into an infinite loop.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
This fixes a bug discovered by generic/388 where sb->s_fs_info was NULL
while the superblock was still active - the error path was entirely
fubar, and was trying to do something unclear and unecessary.
Signed-off-by: Kent Overstreet <kent.overstreet@linux.dev>
