| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
Add ras_poison_irq and functions.
Suggested-by: Hawking Zhang <[email protected]>
Signed-off-by: Horatio Zhang <[email protected]>
Reviewed-by: Tao Zhou <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
Separate vcn RAS poison consumption handling from the instance irq, and
register dedicated ras_poison_irq src and funcs for UVD_POISON.
v2:
- Separate ras irq from vcn instance irq
- Improve the subject and code comments
v3:
- Split the patch into three parts
- Improve the code comments
Suggested-by: Hawking Zhang <[email protected]>
Signed-off-by: Horatio Zhang <[email protected]>
Reviewed-by: Tao Zhou <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This reverts commit 474f01015ffdb74e01c2eb3584a2822c64e7b2be.
Caused a regression:
Samsung Odyssey Neo G9, running at 5120x1440@240/VRR, connected to Navi
21 via DisplayPort, blanks and the GPU hangs while starting the Steam
game Assetto Corsa Competizione (via Proton 7.0).
Example dmesg excerpt:
amdgpu 0000:0c:00.0: [drm] ERROR [CRTC:82:crtc-0] flip_done timed out
NMI watchdog: Watchdog detected hard LOCKUP on cpu 6
[...]
RIP: 0010:amdgpu_device_rreg.part.0+0x2f/0xf0 [amdgpu]
Code: 41 54 44 8d 24 b5 00 00 00 00 55 89 f5 53 48 89 fb 4c 3b a7 60 0b 00 00 73 6a 83 e2 02 74 29 4c 03 a3 68 0b 00 00 45 8b 24 24 <48> 8b 43 08 0f b7 70 3e 66 90 44 89 e0 5b 5d 41 5c 31 d2 31 c9 31
RSP: 0000:ffffb39a119dfb88 EFLAGS: 00000086
RAX: ffffffffc0eb96a0 RBX: ffff9e7963dc0000 RCX: 0000000000007fff
RDX: 0000000000000000 RSI: 0000000000004ff6 RDI: ffff9e7963dc0000
RBP: 0000000000004ff6 R08: ffffb39a119dfc40 R09: 0000000000000010
R10: ffffb39a119dfc40 R11: ffffb39a119dfc44 R12: 00000000000e05ae
R13: 0000000000000000 R14: ffff9e7963dc0010 R15: 0000000000000000
FS: 000000001012f6c0(0000) GS:ffff9e805eb80000(0000) knlGS:000000007fd40000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00000000461ca000 CR3: 00000002a8a20000 CR4: 0000000000350ee0
Call Trace:
<TASK>
dm_read_reg_func+0x37/0xc0 [amdgpu]
generic_reg_get2+0x22/0x60 [amdgpu]
optc1_get_crtc_scanoutpos+0x6a/0xc0 [amdgpu]
dc_stream_get_scanoutpos+0x74/0x90 [amdgpu]
dm_crtc_get_scanoutpos+0x82/0xf0 [amdgpu]
amdgpu_display_get_crtc_scanoutpos+0x91/0x190 [amdgpu]
? dm_read_reg_func+0x37/0xc0 [amdgpu]
amdgpu_get_vblank_counter_kms+0xb4/0x1a0 [amdgpu]
dm_pflip_high_irq+0x213/0x2f0 [amdgpu]
amdgpu_dm_irq_handler+0x8a/0x200 [amdgpu]
amdgpu_irq_dispatch+0xd4/0x220 [amdgpu]
amdgpu_ih_process+0x7f/0x110 [amdgpu]
amdgpu_irq_handler+0x1f/0x70 [amdgpu]
__handle_irq_event_percpu+0x46/0x1b0
handle_irq_event+0x34/0x80
handle_edge_irq+0x9f/0x240
__common_interrupt+0x66/0x110
common_interrupt+0x5c/0xd0
asm_common_interrupt+0x22/0x40
Reviewed-by: Aurabindo Pillai <[email protected]>
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Michel Dänzer <[email protected]>
Signed-off-by: Hamza Mahfooz <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This reverts commit ce560ac40272a5c8b5b68a9d63a75edd9e66aed2.
It depends on its parent commit, which we want to revert.
Acked-by: Alex Deucher <[email protected]>
Signed-off-by: Michel Dänzer <[email protected]>
[Hamza: fix a whitespace issue in dcn30_prepare_bandwidth()]
Signed-off-by: Hamza Mahfooz <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
|
|
This patch reverses the DPM clocks levels output of pp_dpm_mclk
and pp_dpm_fclk for renoir.
On dGPUs and older APUs we expose the levels from lowest clocks
to highest clocks. But for some APUs, the clocks levels are
given the reversed orders by PMFW. Like the memory DPM clocks
that are exposed by pp_dpm_mclk.
It's not intuitive that they are reversed on these APUs. All tools
and software that talks to the driver then has to know different ways
to interpret the data depending on the asic.
So we need to reverse them to expose the clocks levels from the
driver consistently.
Signed-off-by: Tim Huang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
This patch reverses the DPM clocks levels output of pp_dpm_mclk
and pp_dpm_fclk.
On dGPUs and older APUs we expose the levels from lowest clocks
to highest clocks. But for some APUs, the clocks levels that from
the DFPstateTable are given the reversed orders by PMFW. Like the
memory DPM clocks that are exposed by pp_dpm_mclk.
It's not intuitive that they are reversed on these APUs. All tools
and software that talks to the driver then has to know different ways
to interpret the data depending on the asic.
So we need to reverse them to expose the clocks levels from the
driver consistently.
Signed-off-by: Tim Huang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
This patch reverses the DPM clocks levels output of pp_dpm_mclk
and pp_dpm_fclk.
On dGPUs and older APUs we expose the levels from lowest clocks
to highest clocks. But for some APUs, the clocks levels that from
the DFPstateTable are given the reversed orders by PMFW. Like the
memory DPM clocks that are exposed by pp_dpm_mclk.
It's not intuitive that they are reversed on these APUs. All tools
and software that talks to the driver then has to know different ways
to interpret the data depending on the asic.
So we need to reverse them to expose the clocks levels from the
driver consistently.
Signed-off-by: Tim Huang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
This patch reverses the DPM clocks levels output of pp_dpm_mclk.
On dGPUs and older APUs we expose the levels from lowest clocks
to highest clocks. But for some APUs, the clocks levels that from
the DFPstateTable are given the reversed orders by PMFW. Like the
memory DPM clocks that are exposed by pp_dpm_mclk.
It's not intuitive that they are reversed on these APUs. All tools
and software that talks to the driver then has to know different ways
to interpret the data depending on the asic.
So we need to reverse them to expose the clocks levels from the
driver consistently.
Signed-off-by: Tim Huang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
This patch reverses the DPM clocks levels output of pp_dpm_mclk
and pp_dpm_fclk.
On dGPUs and older APUs we expose the levels from lowest clocks
to highest clocks. But for some APUs, the clocks levels that from
the DFPstateTable are given the reversed orders by PMFW. Like the
memory DPM clocks that are exposed by pp_dpm_mclk.
It's not intuitive that they are reversed on these APUs. All tools
and software that talks to the driver then has to know different ways
to interpret the data depending on the asic.
So we need to reverse them to expose the clocks levels from the
driver consistently.
Signed-off-by: Tim Huang <[email protected]>
Reviewed-by: Alex Deucher <[email protected]>
Signed-off-by: Alex Deucher <[email protected]>
Cc: [email protected]
|
|
Until commit 5c2712387d48 ("cacheinfo: Fix LLC is not exported through
sysfs"), cacheinfo called populate_cache_leaves() for CPU coming online
which let the arch specific functions handle (at least on x86)
populating the shared_cpu_map. However, with the changes in the
aforementioned commit, populate_cache_leaves() is not called when a CPU
comes online as a result of hotplug since last_level_cache_is_valid()
returns true as the cacheinfo data is not discarded. The CPU coming
online is not present in shared_cpu_map, however, it will not be added
since the cpu_cacheinfo->cpu_map_populated flag is set (it is set in
populate_cache_leaves() when cacheinfo is first populated for x86)
This can lead to inconsistencies in the shared_cpu_map when an offlined
CPU comes online again. Example below depicts the inconsistency in the
shared_cpu_list in cacheinfo when CPU8 is offlined and onlined again on
a 3rd Generation EPYC processor:
# for i in /sys/devices/system/cpu/cpu8/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu8/cache/index0/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index1/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index2/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index3/shared_cpu_list: 8-15,136-143
# echo 0 > /sys/devices/system/cpu/cpu8/online
# echo 1 > /sys/devices/system/cpu/cpu8/online
# for i in /sys/devices/system/cpu/cpu8/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu8/cache/index0/shared_cpu_list: 8
/sys/devices/system/cpu/cpu8/cache/index1/shared_cpu_list: 8
/sys/devices/system/cpu/cpu8/cache/index2/shared_cpu_list: 8
/sys/devices/system/cpu/cpu8/cache/index3/shared_cpu_list: 8
# cat /sys/devices/system/cpu/cpu136/cache/index0/shared_cpu_list
136
# cat /sys/devices/system/cpu/cpu136/cache/index3/shared_cpu_list
9-15,136-143
Clear the flag when the CPU is removed from shared_cpu_map when
cache_shared_cpu_map_remove() is called during CPU hotplug. This will
allow cache_shared_cpu_map_setup() to add the CPU coming back online in
the shared_cpu_map. Set the flag again when the shared_cpu_map is setup.
Following are results of performing the same test as described above with
the changes:
# for i in /sys/devices/system/cpu/cpu8/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu8/cache/index0/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index1/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index2/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index3/shared_cpu_list: 8-15,136-143
# echo 0 > /sys/devices/system/cpu/cpu8/online
# echo 1 > /sys/devices/system/cpu/cpu8/online
# for i in /sys/devices/system/cpu/cpu8/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu8/cache/index0/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index1/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index2/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index3/shared_cpu_list: 8-15,136-143
# cat /sys/devices/system/cpu/cpu136/cache/index0/shared_cpu_list
8,136
# cat /sys/devices/system/cpu/cpu136/cache/index3/shared_cpu_list
8-15,136-143
Fixes: 5c2712387d48 ("cacheinfo: Fix LLC is not exported through sysfs")
Signed-off-by: K Prateek Nayak <[email protected]>
Reviewed-by: Yicong Yang <[email protected]>
Reviewed-by: Sudeep Holla <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
While building the shared_cpu_map, check if the cache level and cache
type matches. On certain systems that build the cache topology based on
the instance ID, there are cases where the same ID may repeat across
multiple cache levels, leading inaccurate topology.
In event of CPU offlining, the cache_shared_cpu_map_remove() does not
consider if IDs at same level are being compared. As a result, when same
IDs repeat across different cache levels, the CPU going offline is not
removed from all the shared_cpu_map.
Below is the output of cache topology of CPU8 and it's SMT sibling after
CPU8 is offlined on a dual socket 3rd Generation AMD EPYC processor
(2 x 64C/128T) running kernel release v6.3:
# for i in /sys/devices/system/cpu/cpu8/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu8/cache/index0/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index1/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index2/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index3/shared_cpu_list: 8-15,136-143
# echo 0 > /sys/devices/system/cpu/cpu8/online
# for i in /sys/devices/system/cpu/cpu136/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu136/cache/index0/shared_cpu_list: 136
/sys/devices/system/cpu/cpu136/cache/index1/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu136/cache/index2/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu136/cache/index3/shared_cpu_list: 9-15,136-143
CPU8 is removed from index0 (L1i) but remains in the shared_cpu_list of
index1 (L1d) and index2 (L2). Since L1i, L1d, and L2 are shared by the
SMT siblings, and they have the same cache instance ID, CPU 2 is only
removed from the first index with matching ID which is index1 (L1i) in
this case. With this fix, the results are as expected when performing
the same experiment on the same system:
# for i in /sys/devices/system/cpu/cpu8/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu8/cache/index0/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index1/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index2/shared_cpu_list: 8,136
/sys/devices/system/cpu/cpu8/cache/index3/shared_cpu_list: 8-15,136-143
# echo 0 > /sys/devices/system/cpu/cpu8/online
# for i in /sys/devices/system/cpu/cpu136/cache/index*/shared_cpu_list; do echo -n "$i: "; cat $i; done
/sys/devices/system/cpu/cpu136/cache/index0/shared_cpu_list: 136
/sys/devices/system/cpu/cpu136/cache/index1/shared_cpu_list: 136
/sys/devices/system/cpu/cpu136/cache/index2/shared_cpu_list: 136
/sys/devices/system/cpu/cpu136/cache/index3/shared_cpu_list: 9-15,136-143
When rebuilding topology, the same problem appears as
cache_shared_cpu_map_setup() implements a similar logic. Consider the
same 3rd Generation EPYC processor: CPUs in Core 1, that share the L1
and L2 caches, have L1 and L2 instance ID as 1. For all the CPUs on
the second chiplet, the L3 ID is also 1 leading to grouping on CPUs from
Core 1 (1, 17) and the entire second chiplet (8-15, 24-31) as CPUs
sharing one cache domain. This went undetected since x86 processors
depended on arch specific populate_cache_leaves() method to repopulate
the shared_cpus_map when CPU came back online until kernel release
v6.3-rc5.
Fixes: 198102c9103f ("cacheinfo: Fix shared_cpu_map to handle shared caches at different levels")
Signed-off-by: K Prateek Nayak <[email protected]>
Reviewed-by: Sudeep Holla <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The crypto_alloc_shash() function doesn't return NULL, it returns
error pointers. Update the check accordingly.
Fixes: 02fe26f25325 ("firmware_loader: Add debug message with checksum for FW file")
Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: Cezary Rojewski <[email protected]>
Acked-by: Luis Chamberlain <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
There was a bug where this code forgot to unlock the tdev->mutex if the
kzalloc() failed. Fix this issue, by moving the allocation outside the
lock.
Fixes: 2d1e952a2b8e ("mailbox: mailbox-test: Fix potential double-free in mbox_test_message_write()")
Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: Lee Jones <[email protected]>
Signed-off-by: Jassi Brar <[email protected]>
|
|
Pull rdma fixes from Jason Gunthorpe:
- Fix 64K ARM page size support in bnxt_re and efa
- bnxt_re fixes for a memory leak, incorrect error handling and a
remove a bogus FW failure when running on a VF
- Update MAINTAINERS for hns and efa
- Fix two rxe regressions added this merge window in error unwind and
incorrect spinlock primitives
- hns gets a better algorithm for allocating page tables to avoid
running out of resources, and a timeout adjustment
- Fix a text case failure in hns
- Use after free in irdma and fix incorrect construction of a WQE
causing mis-execution
* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
RDMA/irdma: Fix Local Invalidate fencing
RDMA/irdma: Prevent QP use after free
MAINTAINERS: Update maintainer of Amazon EFA driver
RDMA/bnxt_re: Do not enable congestion control on VFs
RDMA/bnxt_re: Fix return value of bnxt_re_process_raw_qp_pkt_rx
RDMA/bnxt_re: Fix a possible memory leak
RDMA/hns: Modify the value of long message loopback slice
RDMA/hns: Fix base address table allocation
RDMA/hns: Fix timeout attr in query qp for HIP08
RDMA/efa: Fix unsupported page sizes in device
RDMA/rxe: Convert spin_{lock_bh,unlock_bh} to spin_{lock_irqsave,unlock_irqrestore}
RDMA/rxe: Fix double unlock in rxe_qp.c
MAINTAINERS: Update maintainers of HiSilicon RoCE
RDMA/bnxt_re: Fix the page_size used during the MR creation
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/maz/arm-platforms into irq/urgent
Pull an irqchip fix from Marc Zyngier:
- Fix regression introduced by the Mediatek workaround.
Link: https://lore.kernel.org/lkml/[email protected]
|
|
No Management involved in Zone Appened.
Fixes: bd83fe6f2cd2 ("nvme: add verbose error logging")
Signed-off-by: Christoph Hellwig <[email protected]>
Reviewed-by: Alan Adamson <[email protected]>
Signed-off-by: Keith Busch <[email protected]>
|
|
gcc 13 may assign another type to enumeration constants than gcc 12. Split
the large enum at the top of source file stex.c such that the type of the
constants used in time expressions is changed back to the same type chosen
by gcc 12. This patch suppresses compiler warnings like this one:
In file included from ./include/linux/bitops.h:7,
from ./include/linux/kernel.h:22,
from drivers/scsi/stex.c:13:
drivers/scsi/stex.c: In function ‘stex_common_handshake’:
./include/linux/typecheck.h:12:25: error: comparison of distinct pointer types lacks a cast [-Werror]
12 | (void)(&__dummy == &__dummy2); \
| ^~
./include/linux/jiffies.h:106:10: note: in expansion of macro ‘typecheck’
106 | typecheck(unsigned long, b) && \
| ^~~~~~~~~
drivers/scsi/stex.c:1035:29: note: in expansion of macro ‘time_after’
1035 | if (time_after(jiffies, before + MU_MAX_DELAY * HZ)) {
| ^~~~~~~~~~
See also https://gcc.gnu.org/bugzilla/show_bug.cgi?id=107405.
Cc: [email protected]
Acked-by: Randy Dunlap <[email protected]>
Tested-by: Randy Dunlap <[email protected]> # build-tested
Signed-off-by: Bart Van Assche <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Martin K. Petersen <[email protected]>
|
|
If an attempt at contacting a receiver or a device fails because the
receiver or device never responds, don't restart the communication, only
restart it if the receiver or device answers that it's busy, as originally
intended.
This was the behaviour on communication timeout before commit 586e8fede795
("HID: logitech-hidpp: Retry commands when device is busy").
This fixes some overly long waits in a critical path on boot, when
checking whether the device is connected by getting its HID++ version.
Signed-off-by: Bastien Nocera <[email protected]>
Suggested-by: Mark Lord <[email protected]>
Fixes: 586e8fede795 ("HID: logitech-hidpp: Retry commands when device is busy")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=217412
Signed-off-by: Jiri Kosina <[email protected]>
|
|
Upon keep alive completion, nvme_keep_alive_work is scheduled with the
same delay every time. If keep alive commands are completing slowly,
this may cause a keep alive timeout. The following trace illustrates the
issue, taking KATO = 8 and TBKAS off for simplicity:
1. t = 0: run nvme_keep_alive_work, send keep alive
2. t = ε: keep alive reaches controller, controller restarts its keep
alive timer
3. t = 4: host receives keep alive completion, schedules
nvme_keep_alive_work with delay 4
4. t = 8: run nvme_keep_alive_work, send keep alive
Here, a keep alive having RTT of 4 causes a delay of at least 8 - ε
between the controller receiving successive keep alives. With ε small,
the controller is likely to detect a keep alive timeout.
Fix this by calculating the RTT of the keep alive command, and adjusting
the scheduling delay of the next keep alive work accordingly.
Reported-by: Costa Sapuntzakis <[email protected]>
Reported-by: Randy Jennings <[email protected]>
Signed-off-by: Uday Shankar <[email protected]>
Reviewed-by: Hannes Reinecke <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Signed-off-by: Keith Busch <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap
Pull regmap fixes from Mark Brown:
"The most important fix here is for missing dropping of the RCU read
lock when syncing maple tree register caches, the physical devices I
have that use the code don't do any syncing so I'd only ever tested
this with virtual devices and missed the fact that we need to drop the
lock in order to write to buses that need to sleep.
Otherwise there's a fix for an edge case when splitting up large batch
writes which has been lurking for a long time, a check to make sure
nobody writes new drivers with a bug that was found in several
SoundWire drivers and a tweak to the way the new kunit tests are
enabled to ensure they don't cause regmap to be enabled when it
wouldn't otherwise be"
* tag 'regmap-fix-v6.4-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regmap:
regmap: maple: Drop the RCU read lock while syncing registers
regmap: sdw: check for invalid multi-register writes config
regmap: Account for register length when chunking
regmap: REGMAP_KUNIT should not select REGMAP
|
|
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.
Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.
Signed-off-by: Uwe Kleine-König <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
|
|
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.
Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.
Signed-off-by: Uwe Kleine-König <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
|
|
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.
Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.
Signed-off-by: Uwe Kleine-König <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
|
|
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.
Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.
Signed-off-by: Uwe Kleine-König <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
|
|
The .remove() callback for a platform driver returns an int which makes
many driver authors wrongly assume it's possible to do error handling by
returning an error code. However the value returned is (mostly) ignored
and this typically results in resource leaks. To improve here there is a
quest to make the remove callback return void. In the first step of this
quest all drivers are converted to .remove_new() which already returns
void.
Trivially convert this driver from always returning zero in the remove
callback to the void returning variant.
Signed-off-by: Uwe Kleine-König <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
|
|
The driver core never calls a remove callback with the platform_device
pointer being NULL. So the check for this condition can just be dropped.
Signed-off-by: Uwe Kleine-König <[email protected]>
Signed-off-by: Helge Deller <[email protected]>
|
|
When a command completes, we set a flag which will skip sending a
keep alive at the next run of nvme_keep_alive_work when TBKAS is on.
However, if the command was submitted long ago, it's possible that
the controller may have also restarted its keep alive timer (as a
result of receiving the command) long ago. The following trace
demonstrates the issue, assuming TBKAS is on and KATO = 8 for
simplicity:
1. t = 0: submit I/O commands A, B, C, D, E
2. t = 0.5: commands A, B, C, D, E reach controller, restart its keep
alive timer
3. t = 1: A completes
4. t = 2: run nvme_keep_alive_work, see recent completion, do nothing
5. t = 3: B completes
6. t = 4: run nvme_keep_alive_work, see recent completion, do nothing
7. t = 5: C completes
8. t = 6: run nvme_keep_alive_work, see recent completion, do nothing
9. t = 7: D completes
10. t = 8: run nvme_keep_alive_work, see recent completion, do nothing
11. t = 9: E completes
At this point, 8.5 seconds have passed without restarting the
controller's keep alive timer, so the controller will detect a keep
alive timeout.
Fix this by checking the IO start time when deciding to defer sending a
keep alive command. Only set comp_seen if the command started after the
most recent run of nvme_keep_alive_work. With this change, the
completions of B, C, and D will not set comp_seen and the run of
nvme_keep_alive_work at t = 4 will send a keep alive.
Reported-by: Costa Sapuntzakis <[email protected]>
Reported-by: Randy Jennings <[email protected]>
Signed-off-by: Uday Shankar <[email protected]>
Reviewed-by: Hannes Reinecke <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Signed-off-by: Keith Busch <[email protected]>
|
|
With TBKAS on, the completion of one command can defer sending a
keep alive for up to twice the delay between successive runs of
nvme_keep_alive_work. The current delay of KATO / 2 thus makes it
possible for one command to defer sending a keep alive for up to
KATO, which can result in the controller detecting a KATO. The following
trace demonstrates the issue, taking KATO = 8 for simplicity:
1. t = 0: run nvme_keep_alive_work, no keep-alive sent
2. t = ε: I/O completion seen, set comp_seen = true
3. t = 4: run nvme_keep_alive_work, see comp_seen == true,
skip sending keep-alive, set comp_seen = false
4. t = 8: run nvme_keep_alive_work, see comp_seen == false,
send a keep-alive command.
Here, there is a delay of 8 - ε between receiving a command completion
and sending the next command. With ε small, the controller is likely to
detect a keep alive timeout.
Fix this by running nvme_keep_alive_work with a delay of KATO / 4
whenever TBKAS is on. Going through the above trace now gives us a
worst-case delay of 4 - ε, which is in line with the recommendation of
sending a command every KATO / 2 in the NVMe specification.
Reported-by: Costa Sapuntzakis <[email protected]>
Reported-by: Randy Jennings <[email protected]>
Signed-off-by: Uday Shankar <[email protected]>
Reviewed-by: Hannes Reinecke <[email protected]>
Reviewed-by: Sagi Grimberg <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Signed-off-by: Keith Busch <[email protected]>
|
|
In the function nvme_passthru_end(), only the value of the command
opcode is checked, without checking the command type (IO command or
Admin command). When we send a Dataset Management command (The opcode
of the Dataset Management command is the same as the Set Feature
command), kernel thinks it is a set feature command, then sets the
controller's keep alive interval, and calls nvme_keep_alive_work().
Signed-off-by: min15.li <[email protected]>
Reviewed-by: Kanchan Joshi <[email protected]>
Reviewed-by: Christoph Hellwig <[email protected]>
Signed-off-by: Keith Busch <[email protected]>
|
|
During system resume we need to resume the polling workqueue
if client->irq is not set else polling will no longer work.
Fixes: 0d6a119cecd7 ("usb: typec: tps6598x: Add support for polling interrupts status")
Signed-off-by: Roger Quadros <[email protected]>
Reviewed-by: Heikki Krogerus <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
In a COMPILE_TEST configuration, the cpm_uart driver uses symbols from
the cpm_uart_cpm2.c file. This file is compiled only when CONFIG_CPM2 is
set.
Without this dependency, the linker fails with some missing symbols for
COMPILE_TEST configuration that needs SERIAL_CPM without enabling CPM2.
This lead to:
depends on CPM2 || CPM1 || (PPC32 && CPM2 && COMPILE_TEST)
This dependency does not make sense anymore and can be simplified
removing all the COMPILE_TEST part.
Signed-off-by: Herve Codina <[email protected]>
Reported-by: kernel test robot <[email protected]>
Link: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Fixes: e3e7b13bffae ("serial: allow COMPILE_TEST for some drivers")
Cc: stable <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
In order to compile tsa.c and qmc.c, CONFIG_CPM must be set.
Without this dependency, the linker fails with some missing
symbols for COMPILE_TEST configurations that need QMC without
enabling CPM.
Signed-off-by: Herve Codina <[email protected]>
Reported-by: kernel test robot <[email protected]>
Link: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Acked-by: Randy Dunlap <[email protected]>
Tested-by: Randy Dunlap <[email protected]> # build-tested
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
UARTCTRL_SBK
LPUART IP now has two known bugs, one is that CTS has higher priority
than the break signal, which causes the break signal sending through
UARTCTRL_SBK may impacted by the CTS input if the HW flow control is
enabled. It exists on all platforms we support in this driver.
So we add a workaround patch for this issue: commit c4c81db5cf8b
("tty: serial: fsl_lpuart: disable the CTS when send break signal").
Another IP bug is i.MX8QM LPUART may have an additional break character
being sent after SBK was cleared. It may need to add some delay between
clearing SBK and re-enabling CTS to ensure that the SBK latch are
completely cleared.
But we found that during the delay period before CTS is enabled, there
is still a risk that Bluetooth data in TX FIFO may be sent out during
this period because of break off and CTS disabled(even if BT sets CTS
line deasserted, data is still sent to BT).
Due to this risk, we have to drop the CTS-disabling workaround for SBK
bugs, use TXINV seems to be a better way to replace SBK feature and
avoid above risk. Also need to disable the transmitter to prevent any
data from being sent out during break, then invert the TX line to send
break. Then disable the TXINV when turn off break and re-enable
transmitter.
Fixes: c4c81db5cf8b ("tty: serial: fsl_lpuart: disable the CTS when send break signal")
Cc: stable <[email protected]>
Signed-off-by: Sherry Sun <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
If an error occurs after reset_control_deassert(), it must be re-asserted,
as already done in the .remove() function.
Fixes: c6825c6395b7 ("serial: 8250_tegra: Create Tegra specific 8250 driver")
Cc: stable <[email protected]>
Signed-off-by: Christophe JAILLET <[email protected]>
Link: https://lore.kernel.org/r/f8130f35339cc80edc6b9aac4bb2a60b60a226bf.1684063511.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The apc->eth_stats.rx_cqes is one per NIC (vport), and it's on the
frequent and parallel code path of all queues. So, r/w into this
single shared variable by many threads on different CPUs creates a
lot caching and memory overhead, hence perf regression. And, it's
not accurate due to the high volume concurrent r/w.
For example, a workload is iperf with 128 threads, and with RPS
enabled. We saw perf regression of 25% with the previous patch
adding the counters. And this patch eliminates the regression.
Since the error path of mana_poll_rx_cq() already has warnings, so
keeping the counter and convert it to a per-queue variable is not
necessary. So, just remove this counter from this high frequency
code path.
Also, remove the tx_cqes counter for the same reason. We have
warnings & other counters for errors on that path, and don't need
to count every normal cqe processing.
Cc: [email protected]
Fixes: bd7fc6e1957c ("net: mana: Add new MANA VF performance counters for easier troubleshooting")
Signed-off-by: Haiyang Zhang <[email protected]>
Reviewed-by: Horatiu Vultur <[email protected]>
Reviewed-by: Jiri Pirko <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Paolo Abeni <[email protected]>
|
|
When checking for OF quirks, make sure either 'compatible' or 'property'
is set, and give up otherwise.
This avoids non-OF quirks being randomly applied as they don't have any
of the OF data that need checking.
Cc: Douglas Anderson <[email protected]>
Reported-by: Geert Uytterhoeven <[email protected]>
Tested-by: Geert Uytterhoeven <[email protected]>
Fixes: 44bd78dd2b88 ("irqchip/gic-v3: Disable pseudo NMIs on Mediatek devices w/ firmware issues")
Signed-off-by: Marc Zyngier <[email protected]>
|
|
BM818 is based on Qualcomm MDM9607 chipset.
Fixes: 9a07406b00cd ("net: usb: qmi_wwan: Add the BroadMobi BM818 card")
Cc: [email protected]
Signed-off-by: Sebastian Krzyszkowiak <[email protected]>
Acked-by: Bjørn Mork <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>
|
|
For devices not attached to a port multiplier and managed directly by
libata, the device number passed to ata_find_dev() must always be lower
than the maximum number of devices returned by ata_link_max_devices().
That is 1 for SATA devices or 2 for an IDE link with master+slave
devices. This device number is the SCSI device ID which matches these
constraints as the IDs are generated per port and so never exceed the
maximum number of devices for the link being used.
However, for libsas managed devices, SCSI device IDs are assigned per
struct scsi_host, leading to device IDs for SATA devices that can be
well in excess of libata per-link maximum number of devices. This
results in ata_find_dev() to always return NULL for libsas managed
devices except for the first device of the target scsi_host with ID
(device number) equal to 0. This issue is visible by executing the
hdparm utility, which fails. E.g.:
hdparm -i /dev/sdX
/dev/sdX:
HDIO_GET_IDENTITY failed: No message of desired type
Fix this by rewriting ata_find_dev() to ignore the device number for
non-PMP attached devices with a link with at most 1 device, that is SATA
devices. For these, the device number 0 is always used to
return the correct pointer to the struct ata_device of the port link.
This change excludes IDE master/slave setups (maximum number of devices
per link is 2) and port-multiplier attached devices. Also, to be
consistant with the fact that SCSI device IDs and channel numbers used
as device numbers are both unsigned int, change the devno argument of
ata_find_dev() to unsigned int.
Reported-by: Xingui Yang <[email protected]>
Fixes: 41bda9c98035 ("libata-link: update hotplug to handle PMP links")
Cc: [email protected]
Signed-off-by: Damien Le Moal <[email protected]>
Reviewed-by: Jason Yan <[email protected]>
|
|
If the local invalidate fence is indicated in the WR, only the read fence
is currently being set in WQE. Fix this to set both the read and local
fence in the WQE.
Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs")
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Mustafa Ismail <[email protected]>
Signed-off-by: Shiraz Saleem <[email protected]>
Signed-off-by: Jason Gunthorpe <[email protected]>
|
|
There is a window where the poll cq may use a QP that has been freed.
This can happen if a CQE is polled before irdma_clean_cqes() can clear the
CQE's related to the QP and the destroy QP races to free the QP memory.
then the QP structures are used in irdma_poll_cq. Fix this by moving the
clearing of CQE's before the reference is removed and the QP is destroyed.
Fixes: b48c24c2d710 ("RDMA/irdma: Implement device supported verb APIs")
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Mustafa Ismail <[email protected]>
Signed-off-by: Shiraz Saleem <[email protected]>
Signed-off-by: Jason Gunthorpe <[email protected]>
|
|
When hcd->localmem_pool is non-null, localmem_pool is used to allocate
DMA memory. In this case, the dma address will be properly returned (in
dma_handle), and dma_mmap_coherent should be used to map this memory
into the user space. However, the current implementation uses
pfn_remap_range, which is supposed to map normal pages.
Instead of repeating the logic in the memory allocation function, this
patch introduces a more robust solution. Here, the type of allocated
memory is checked by testing whether dma_handle is properly set. If
dma_handle is properly returned, it means some DMA pages are allocated
and dma_mmap_coherent should be used to map them. Otherwise, normal
pages are allocated and pfn_remap_range should be called. This ensures
that the correct mmap functions are used consistently, independently
with logic details that determine which type of memory gets allocated.
Fixes: a0e710a7def4 ("USB: usbfs: fix mmap dma mismatch")
Cc: [email protected]
Signed-off-by: Ruihan Li <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The current implementation of usbdev_mmap uses usb_alloc_coherent to
allocate memory pages that will later be mapped into the user space.
Meanwhile, usb_alloc_coherent employs three different methods to
allocate memory, as outlined below:
* If hcd->localmem_pool is non-null, it uses gen_pool_dma_alloc to
allocate memory;
* If DMA is not available, it uses kmalloc to allocate memory;
* Otherwise, it uses dma_alloc_coherent.
However, it should be noted that gen_pool_dma_alloc does not guarantee
that the resulting memory will be page-aligned. Furthermore, trying to
map slab pages (i.e., memory allocated by kmalloc) into the user space
is not resonable and can lead to problems, such as a type confusion bug
when PAGE_TABLE_CHECK=y [1].
To address these issues, this patch introduces hcd_alloc_coherent_pages,
which addresses the above two problems. Specifically,
hcd_alloc_coherent_pages uses gen_pool_dma_alloc_align instead of
gen_pool_dma_alloc to ensure that the memory is page-aligned. To replace
kmalloc, hcd_alloc_coherent_pages directly allocates pages by calling
__get_free_pages.
Reported-by: [email protected]
Closes: https://lore.kernel.org/lkml/[email protected]/ [1]
Fixes: f7d34b445abc ("USB: Add support for usbfs zerocopy.")
Fixes: ff2437befd8f ("usb: host: Fix excessive alignment restriction for local memory allocations")
Cc: [email protected]
Signed-off-by: Ruihan Li <[email protected]>
Acked-by: Alan Stern <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The "udc" pointer was never set in the probe() function so it will
lead to a NULL dereference in udc_pci_remove() when we do:
usb_del_gadget_udc(&udc->gadget);
Signed-off-by: Dan Carpenter <[email protected]>
Link: https://lore.kernel.org/r/ZG+A/dNpFWAlCChk@kili
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
While exercising the unbind path, with the current implementation
the functionfs_unbind would be calling which waits for the ffs->mutex
to be available, however within the same time ffs_ep0_read is invoked
& if no setup packets are pending, it will invoke function
wait_event_interruptible_exclusive_locked_irq which by definition waits
for the ev.count to be increased inside the same mutex for which
functionfs_unbind is waiting.
This creates deadlock situation because the functionfs_unbind won't
get the lock until ev.count is increased which can only happen if
the caller ffs_func_unbind can proceed further.
Following is the illustration:
CPU1 CPU2
ffs_func_unbind() ffs_ep0_read()
mutex_lock(ffs->mutex)
wait_event(ffs->ev.count)
functionfs_unbind()
mutex_lock(ffs->mutex)
mutex_unlock(ffs->mutex)
ffs_event_add()
<deadlock>
Fix this by moving the event unbind before functionfs_unbind
to ensure the ev.count is incrased properly.
Fixes: 6a19da111057 ("usb: gadget: f_fs: Prevent race during ffs_ep0_queue_wait")
Cc: stable <[email protected]>
Signed-off-by: Uttkarsh Aggarwal <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
At iMX8QM platform, enable NCM gadget and run 'iperf3 -s'.
At host, run 'iperf3 -V -c fe80::6863:98ff:feef:3e0%enxc6e147509498'
[ 5] 0.00-1.00 sec 1.55 MBytes 13.0 Mbits/sec 90 4.18 KBytes
[ 5] 1.00-2.00 sec 1.44 MBytes 12.0 Mbits/sec 75 4.18 KBytes
[ 5] 2.00-3.00 sec 1.48 MBytes 12.4 Mbits/sec 75 4.18 KBytes
Expected speed should be bigger than 300Mbits/sec.
The root cause of this performance drop was found to be data corruption
happening at 4K borders in some Ethernet packets, leading to TCP
checksum errors. This corruption occurs from the position
(4K - (address & 0x7F)) to 4K. The u_ether function's allocation of
skb_buff reserves 64B, meaning all RX addresses resemble 0xXXXX0040.
Force trb_burst_size to 16 can fix this problem.
Cc: [email protected]
Fixes: 7733f6c32e36 ("usb: cdns3: Add Cadence USB3 DRD Driver")
Signed-off-by: Frank Li <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The channel's rpmsg object allows new invocations to be made. After old
invocations are already interrupted, the driver shouldn't try to invoke
anymore. Invalidating the rpmsg at the end of the driver removal
function makes it easy to cause a race condition in userspace. Even
closing a file descriptor before the driver finishes its cleanup can
cause an invocation via fastrpc_release_current_dsp_process() and
subsequent timeout.
Invalidate the channel before the invocations are interrupted to make
sure that no invocations can be created to hang after the device closes.
Fixes: c68cfb718c8f ("misc: fastrpc: Add support for context Invoke method")
Cc: stable <[email protected]>
Signed-off-by: Richard Acayan <[email protected]>
Signed-off-by: Srinivas Kandagatla <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The return value is initialized as -1, or -EPERM. The completion of an
invocation implies that the return value is set appropriately, but
"Permission denied" does not accurately describe the outcome of the
invocation. Set the invocation's return value to a more appropriate
"Broken pipe", as the cleanup breaks the driver's connection with rpmsg.
Fixes: c68cfb718c8f ("misc: fastrpc: Add support for context Invoke method")
Cc: stable <[email protected]>
Signed-off-by: Richard Acayan <[email protected]>
Reviewed-by: Srinivas Kandagatla <[email protected]>
Signed-off-by: Srinivas Kandagatla <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
The userspace map request for remote heap allocates CMA memory.
The ownership of this memory needs to be reassigned to proper
owners to allow access from the protection domain running on
DSP. This reassigning of ownership is not correct if done for
any other supported flags.
When any other flag is requested from userspace, fastrpc is
trying to reassign the ownership of memory and this reassignment
is getting skipped for remote heap request which is incorrect.
Add proper flag check to reassign the memory only if remote heap
is requested.
Fixes: 532ad70c6d44 ("misc: fastrpc: Add mmap request assigning for static PD pool")
Cc: stable <[email protected]>
Tested-by: Ekansh Gupta <[email protected]>
Signed-off-by: Ekansh Gupta <[email protected]>
Reviewed-by: Srinivas Kandagatla <[email protected]>
Signed-off-by: Srinivas Kandagatla <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
If a map request is made with securemap attribute, the memory
ownership needs to be reassigned to new VMID to allow access
from protection domain. Currently only DSP VMID is passed to
the reassign call which is incorrect as only a combination of
HLOS and DSP VMID is allowed for memory ownership reassignment
and passing only DSP VMID will cause assign call failure.
Also pass proper restoring permissions to HLOS as the source
permission will now carry both HLOS and DSP VMID permission.
Change is also made to get valid physical address from
scatter/gather for this allocation request.
Fixes: e90d91190619 ("misc: fastrpc: Add support to secure memory map")
Cc: stable <[email protected]>
Tested-by: Ekansh Gupta <[email protected]>
Signed-off-by: Ekansh Gupta <[email protected]>
Signed-off-by: Srinivas Kandagatla <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Greg Kroah-Hartman <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy
Pull phy fixes from Vinod Koul:
- init count imbalance fix in qcom-qmp-pcie and combo drivers
- kernel doc header fix for qcom-snps driver
- mediatek floating point comparison fix
- amlogic fix register value
* tag 'phy-fixes-6.4' of git://git.kernel.org/pub/scm/linux/kernel/git/phy/linux-phy:
phy: qcom-snps: correct struct qcom_snps_hsphy kerneldoc
phy: amlogic: phy-meson-g12a-mipi-dphy-analog: fix CNTL2_DIF_TX_CTL0 value
phy: mediatek: rework the floating point comparisons to fixed point
phy: qcom-qmp-pcie-msm8996: fix init-count imbalance
phy: qcom-qmp-combo: fix init-count imbalance
|