When unloading the module, one gets
------------[ cut here ]------------
Device 'cmm0' does not have a release() function, it is broken and must be fixed. See Documentation/kobject.txt.
WARNING: CPU: 0 PID: 19308 at drivers/base/core.c:1244 .device_release+0xcc/0xf0
...
We only have one static fake device. There is nothing to do when
releasing the device (via cmm_exit()).
Signed-off-by: David Hildenbrand <david@redhat.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20191031142933.10779-2-david@redhat.com Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
In function __ufshcd_query_descriptor(), in the event of an error
happening, we directly goto out_unlock and forget to invaliate
hba->dev_cmd.query.descriptor pointer. This results in this pointer still
valid in ufshcd_copy_query_response() for other query requests which go
through ufshcd_exec_raw_upiu_cmd(). This will cause __memcpy() crash and
system hangs. Log as shown below:
Unable to handle kernel paging request at virtual address ffff000012233c40
Mem abort info:
ESR = 0x96000047
Exception class = DABT (current EL), IL = 32 bits
SET = 0, FnV = 0
EA = 0, S1PTW = 0
Data abort info:
ISV = 0, ISS = 0x00000047
CM = 0, WnR = 1
swapper pgtable: 4k pages, 48-bit VAs, pgdp = 0000000028cc735c
[ffff000012233c40] pgd=00000000bffff003, pud=00000000bfffe003,
pmd=00000000ba8b8003, pte=0000000000000000
Internal error: Oops: 96000047 [#2] PREEMPT SMP
...
Call trace:
__memcpy+0x74/0x180
ufshcd_issue_devman_upiu_cmd+0x250/0x3c0
ufshcd_exec_raw_upiu_cmd+0xfc/0x1a8
ufs_bsg_request+0x178/0x3b0
bsg_queue_rq+0xc0/0x118
blk_mq_dispatch_rq_list+0xb0/0x538
blk_mq_sched_dispatch_requests+0x18c/0x1d8
__blk_mq_run_hw_queue+0xb4/0x118
blk_mq_run_work_fn+0x28/0x38
process_one_work+0x1ec/0x470
worker_thread+0x48/0x458
kthread+0x130/0x138
ret_from_fork+0x10/0x1c
Code: 540000aba8c12027a88120c7a8c12027 (a88120c7)
---[ end trace 793e1eb5dff69f2d ]---
note: kworker/0:2H[2054] exited with preempt_count 1
This patch is to move "descriptor = NULL" down to below the label
"out_unlock".
Fixes: d44a5f98bb49b2(ufs: query descriptor API) Link: https://lore.kernel.org/r/20191112223436.27449-3-huobean@gmail.com Reviewed-by: Alim Akhtar <alim.akhtar@samsung.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> Signed-off-by: Bean Huo <beanhuo@micron.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
The first entry of the ibm,drc-info property is an int encoded count
of the number of drc-info entries that follow. The "value" pointer
returned by of_prop_next_u32() is still pointing at the this value
when we call of_read_drc_info_cell(), but the helper function
expects that value to be pointing at the first element of an entry.
Fix up by incrementing the "value" pointer to point at the first
element of the first drc-info entry prior.
When using this driver on a Blizzard 1260, there were failures whenever DMA
transfers from the SCSI bus to memory of 65535 bytes were followed by a DMA
transfer of 1 byte. This caused the byte at offset 65535 to be overwritten
with 0xff. The Blizzard hardware can't handle single byte DMA transfers.
Besides this issue, limiting the DMA length to something that is not a
multiple of the page size is very inefficient on most file systems.
It seems this limit was chosen because the DMA transfer counter of the ESP
by default is 16 bits wide, thus limiting the length to 65535 bytes.
However, the value 0 means 65536 bytes, which is handled by the ESP and the
Blizzard just fine. It is also the default maximum used by esp_scsi when
drivers don't provide their own dma_length_limit() function.
The limit of 65536 bytes can be used by all boards except the Fastlane. The
old driver used a limit of 65532 bytes (0xfffc), which is reintroduced in
this patch.
Fixes: b7ded0e8b0d1 ("scsi: zorro_esp: Limit DMA transfers to 65535 bytes") Link: https://lore.kernel.org/r/20191112175523.23145-1-jongk@linux-m68k.org Signed-off-by: Kars de Jong <jongk@linux-m68k.org> Reviewed-by: Finn Thain <fthain@telegraphics.com.au> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
*** CID 101747: Null pointer dereferences (FORWARD_NULL)
/drivers/scsi/lpfc/lpfc_els.c: 4439 in lpfc_cmpl_els_rsp()
4433 kfree(mp);
4434 }
4435 mempool_free(mbox, phba->mbox_mem_pool);
4436 }
4437 out:
4438 if (ndlp && NLP_CHK_NODE_ACT(ndlp)) {
vvv CID 101747: Null pointer dereferences (FORWARD_NULL)
vvv Dereferencing null pointer "shost".
4439 spin_lock_irq(shost->host_lock);
4440 ndlp->nlp_flag &= ~(NLP_ACC_REGLOGIN | NLP_RM_DFLT_RPI);
4441 spin_unlock_irq(shost->host_lock);
4442
4443 /* If the node is not being used by another discovery thread,
4444 * and we are sending a reject, we are done with it.
Fix by adding a check for non-null shost in line 4438.
The scenario when shost is set to null is when ndlp is null.
As such, the ndlp check present was sufficient. But better safe
than sorry so add the shost check.
Reported-by: coverity-bot <keescook+coverity-bot@chromium.org>
Addresses-Coverity-ID: 101747 ("Null pointer dereferences") Fixes: 2e0fef85e098 ("[SCSI] lpfc: NPIV: split ports") CC: James Bottomley <James.Bottomley@SteelEye.com> CC: "Gustavo A. R. Silva" <gustavo@embeddedor.com> CC: linux-next@vger.kernel.org Link: https://lore.kernel.org/r/20191111230401.12958-3-jsmart2021@gmail.com Reviewed-by: Ewan D. Milne <emilne@redhat.com> Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Naresh reported LTP diotest4 failing for 32bit x86 and arm -next
kernels on ext4. Same problem exists in 5.4-rc7 on xfs.
The failure comes down to:
openat(AT_FDCWD, "testdata-4.5918", O_RDWR|O_DIRECT) = 4
mmap2(NULL, 4096, PROT_READ, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0xb7f7b000
read(4, 0xb7f7b000, 4096) = 0 // expects -EFAULT
Problem is conversion at iomap_dio_bio_actor() return. Ternary
operator has a return type and an attempt is made to convert each
of operands to the type of the other. In this case "ret" (int)
is converted to type of "copied" (unsigned long). Both have size
of 4 bytes:
size_t copied = 0;
int ret = -14;
long long actor_ret = copied ? copied : ret;
On x86_64: actor_ret == -14;
On x86 : actor_ret == 4294967282
Replace ternary operator with 2 return statements to avoid this
unwanted conversion.
Fixes: 4721a6010990 ("iomap: dio data corruption and spurious errors when pipes fill") Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org> Signed-off-by: Jan Stancek <jstancek@redhat.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Add missing dma channels free calls in case of error during probe
and reorder the remove function so that dma channels are freed after
the i2c adapter is deleted.
Overall, reorder the remove function so that probe error handling order
and remove function order are same.
Since commit 7723f4c5ecdb ("driver core: platform: Add an error message
to platform_get_irq*()"), platform_get_irq_byname() displays an error
when the IRQ isn't found. Since the SMMUv3 driver uses that function to
query which interrupt method is available, the message is now displayed
during boot for any SMMUv3 that doesn't implement the combined
interrupt, or that implements MSIs.
[ 20.700337] arm-smmu-v3 arm-smmu-v3.7.auto: IRQ combined not found
[ 20.706508] arm-smmu-v3 arm-smmu-v3.7.auto: IRQ eventq not found
[ 20.712503] arm-smmu-v3 arm-smmu-v3.7.auto: IRQ priq not found
[ 20.718325] arm-smmu-v3 arm-smmu-v3.7.auto: IRQ gerror not found
Use platform_get_irq_byname_optional() to avoid displaying a spurious
error.
Quota statistics counted as 64-bit per-cpu counter. Reading sums per-cpu
fractions as signed 64-bit int, filters negative values and then reports
lower half as signed 32-bit int.
As seen on the new Raspberry Pi 4 and sta2x11's DMA implementation it is
possible for a device configured with 32 bit DMA addresses and a partial
DMA mapping located at the end of the address space to overflow. It
happens when a higher physical address, not DMAable, is translated to
it's DMA counterpart.
For example the Raspberry Pi 4, configurable up to 4 GB of memory, has
an interconnect capable of addressing the lower 1 GB of physical memory
with a DMA offset of 0xc0000000. It transpires that, any attempt to
translate physical addresses higher than the first GB will result in an
overflow which dma_capable() can't detect as it only checks for
addresses bigger then the maximum allowed DMA address.
Fix this by verifying in dma_capable() if the DMA address range provided
is at any point lower than the minimum possible DMA address on the bus.
Signed-off-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de> Signed-off-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
For an external clock source, which is gated via a GPIO, the
rate change should typically be propagated to the parent clock.
The situation where we are requiring this propagation, is when an
external clock is connected to override an internal clock (which typically
has a fixed rate). The external clock can have a different rate than the
internal one, and may also be variable, thus requiring the rate
propagation.
This rate change wasn't propagated until now, and it's unclear about cases
where this shouldn't be propagated. Thus, it's unclear whether this is
fixing a bug, or extending the current driver behavior. Also, it's unsure
about whether this may break any existing setups; in the case that it does,
a device-tree property may be added to disable this flag.
Signed-off-by: Michael Hennerich <michael.hennerich@analog.com> Signed-off-by: Alexandru Ardelean <alexandru.ardelean@analog.com> Link: https://lkml.kernel.org/r/20191108071718.17985-1-alexandru.ardelean@analog.com Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Some RCGs (the gfx_3d_src_clk in msm8998 for example) are basically just
some constant ratio from the input across the entire frequency range. It
would be great if we could specify the frequency table as a single entry
constant ratio instead of a long list, ie:
{ .src = P_GPUPLL0_OUT_EVEN, .pre_div = 3 },
{ }
So, lets support that.
We need to fix a corner case in qcom_find_freq() where if the freq table
is non-null, but has no frequencies, we end up returning an "entry" before
the table array, which is bad. Then, we need ignore the freq from the
table, and instead base everything on the requested freq.
Suggested-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Jeffrey Hugo <jeffrey.l.hugo@gmail.com> Link: https://lkml.kernel.org/r/20191031185715.15504-1-jeffrey.l.hugo@gmail.com Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
When MSM8998 support was added, and analysis was done to determine what
clocks would be consumed. That analysis had a flaw, which caused the
pnoc to be skipped. The pnoc clock needs to be on to access the uart
for the console. The clock is on from boot, but has no consumer votes
in the RPM. When we attempt to boot the modem, it causes the RPM to
turn off pnoc, which kills our access to the console and causes CPU hangs.
We need pnoc to be defined, so that clk_smd_rpm_handoff() will put in
an implicit vote for linux and prevent issues when booting modem.
Hopefully pnoc can be consumed by the interconnect framework in future
so that Linux can rely on explicit votes.
Fixes: 6131dc81211c ("clk: qcom: smd: Add support for MSM8998 rpm clocks") Signed-off-by: Jeffrey Hugo <jeffrey.l.hugo@gmail.com> Link: https://lkml.kernel.org/r/20191107190615.5656-1-jeffrey.l.hugo@gmail.com Signed-off-by: Stephen Boyd <sboyd@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
This is causing xfstest generic/579 to fail due to fsck.f2fs reporting errors.
I'm not sure what the problem is, but it still happens even with all the
fs-verity stuff in the test commented out, so that the test just runs fsstress.
generic/579 23s ... [10:02:25]
[ 7.745370] run fstests generic/579 at 2019-11-04 10:02:25
_check_generic_filesystem: filesystem on /dev/vdc is inconsistent
(see /results/f2fs/results-default/generic/579.full for details)
[10:02:47]
Ran: generic/579
Failures: generic/579
Failed 1 of 1 tests
Xunit report: /results/f2fs/results-default/result.xml
Here's the contents of 579.full:
_check_generic_filesystem: filesystem on /dev/vdc is inconsistent
*** fsck.f2fs output ***
[ASSERT] (__chk_dots_dentries:1378) --> Bad inode number[0x24] for '..', parent parent ino is [0xd10]
The root cause is that we forgot to update directory's i_pino during
cross_rename, fix it.
Fixes: 32f9bc25cbda0 ("f2fs: support ->rename2()") Signed-off-by: Chao Yu <yuchao0@huawei.com> Tested-by: Eric Biggers <ebiggers@kernel.org> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
If the driver receives a login that is later then LOGO'd by the remote port
(aka ndlp), the driver, upon the completion of the LOGO ACC transmission,
will logout the node and unregister the rpi that is being used for the
node. As part of the unreg, the node's rpi value is replaced by the
LPFC_RPI_ALLOC_ERROR value. If the port is subsequently offlined, the
offline walks the nodes and ensures they are logged out, which possibly
entails unreg'ing their rpi values. This path does not validate the node's
rpi value, thus doesn't detect that it has been unreg'd already. The
replaced rpi value is then used when accessing the rpi bitmask array which
tracks active rpi values. As the LPFC_RPI_ALLOC_ERROR value is not a valid
index for the bitmask, it may fault the system.
Revise the rpi release code to detect when the rpi value is the replaced
RPI_ALLOC_ERROR value and ignore further release steps.
Link: https://lore.kernel.org/r/20191105005708.7399-2-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
During heavy RCN activity and log_verbose = 0 we see these messages:
2754 PRLI failure DID:521245 Status:x9/xb2c00, data: x0
0231 RSCN timeout Data: x0 x3
0230 Unexpected timeout, hba link state x5
This is due to delayed RSCN activity.
Correct by avoiding the timeout thus the messages by restarting the
discovery timeout whenever an rscn is received.
Filter PRLI responses such that severity depends on whether expected for
the configuration or not. For example, PRLI errors on a fabric will be
informational (they are expected), but Point-to-Point errors are not
necessarily expected so they are raised to an error level.
Link: https://lore.kernel.org/r/20191105005708.7399-5-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
jbd2 statistics counting number of blocks logged in a transaction was
wrong. It didn't count the commit block and more importantly it didn't
count revoke descriptor blocks. Make sure these get properly counted.
This patch addresses what Dave Chinner had discovered and fixed within
commit: 7684e2c4384d. This changes does not have any user visible
impact for ext4 as none of the current users of ext4_iomap_begin()
that extend files depend on IOMAP_F_DIRTY.
When doing a direct IO that spans the current EOF, and there are
written blocks beyond EOF that extend beyond the current write, the
only metadata update that needs to be done is a file size extension.
However, we don't mark such iomaps as IOMAP_F_DIRTY to indicate that
there is IO completion metadata updates required, and hence we may
fail to correctly sync file size extensions made in IO completion when
O_DSYNC writes are being used and the hardware supports FUA.
Hence when setting IOMAP_F_DIRTY, we need to also take into account
whether the iomap spans the current EOF. If it does, then we need to
mark it dirty so that IO completion will call generic_write_sync() to
flush the inode size update to stable storage correctly.
This patch updates the lock pattern in ext4_direct_IO_read() to not
block on inode lock in cases of IOCB_NOWAIT direct I/O reads. The
locking condition implemented here is similar to that of 942491c9e6d6
("xfs: fix AIM7 regression").
Fixes: 16c54688592c ("ext4: Allow parallel DIO reads") Signed-off-by: Matthew Bobrowski <mbobrowski@mbobrowski.org> Reviewed-by: Jan Kara <jack@suse.cz> Reviewed-by: Ritesh Harjani <riteshh@linux.ibm.com> Link: https://lore.kernel.org/r/c5d5e759f91747359fbd2c6f9a36240cf75ad79f.1572949325.git.mbobrowski@mbobrowski.org Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
With large memory (8TB and more) hotplug, we can get soft lockup
warnings as below. These were caused by a long loop without any
explicit cond_resched which is a problem for !PREEMPT kernels.
Avoid this using cond_resched() while inserting hash page table
entries. We already do similar cond_resched() in __add_pages(), see
commit f64ac5e6e306 ("mm, memory_hotplug: add scheduling point to
__add_pages").
Using Makefile's wildcard with absolute path to detect
the presence of libyaml results in false-positive
detection when cross-compiling e.g. in yocto environment.
The latter results in build error:
| scripts/dtc/yamltree.o: In function `yaml_propval_int':
| yamltree.c: undefined reference to `yaml_sequence_start_event_initialize'
| yamltree.c: undefined reference to `yaml_emitter_emit'
| yamltree.c: undefined reference to `yaml_scalar_event_initialize'
...
Use pkg-config to locate libyaml to address this scenario.
Signed-off-by: Pavel Modilaynen <pavel.modilaynen@axis.com>
[robh: silence stderr] Signed-off-by: Rob Herring <robh@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
If a hardware-specific driver does not provide a name, the timer-of core
falls back to device_node.name. Due to generic DT node naming policies,
that name is almost always "timer", and thus doesn't identify the actual
timer used.
Fix this by using device_node.full_name instead, which includes the unit
addrees.
The NETDEV_CHANGENAME code is not "unneeded" like it is stated in commit 4cb6560514fa ("leds: trigger: netdev: fix refcnt leak on interface
rename").
The event was accidentally misinterpreted equivalent to
NETDEV_UNREGISTER, but should be equivalent to NETDEV_REGISTER.
This was the case in the original code from the openwrt project.
Otherwise, you are unable to set netdev led triggers for (non-existent)
netdevices, which has to be renamed. This is the case, for example, for
ppp interfaces in openwrt.
Fixes: 06f502f57d0d ("leds: trigger: Introduce a NETDEV trigger") Fixes: 4cb6560514fa ("leds: trigger: netdev: fix refcnt leak on interface rename") Signed-off-by: Martin Schiller <ms@dev.tdt.de> Signed-off-by: Pavel Machek <pavel@ucw.cz> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
platform_get_irq_byname() might return -errno which later would be cast
to an unsigned int and used in IRQ handling code leading to usage of
wrong ID and errors about wrong irq_base.
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org> Reviewed-by: Peng Ma <peng.ma@nxp.com> Tested-by: Peng Ma <peng.ma@nxp.com> Link: https://lore.kernel.org/r/20191004150826.6656-1-krzk@kernel.org Signed-off-by: Vinod Koul <vkoul@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Daniele reported that issue previously fixed in c41f9ea998f3
("drivers: dma-coherent: Account dma_pfn_offset when used with device
tree") reappear shortly after 43fc509c3efb ("dma-coherent: introduce
interface for default DMA pool") where fix was accidentally dropped.
Lets put fix back in place and respect dma-ranges for reserved memory.
As we've seen from USB and other areas[1], we need to always do runtime
checks for DMA operating on memory regions that might be remapped. This
adds vmap checks (similar to those already in USB but missing in other
places) into dma_map_single() so all callers benefit from the checking.
Some of our scripts are passed $objdump and then call it as
"$objdump". This doesn't work if it contains spaces because we're
using ccache, for example you get errors such as:
./arch/powerpc/tools/relocs_check.sh: line 48: ccache ppc64le-objdump: No such file or directory
./arch/powerpc/tools/unrel_branch_check.sh: line 26: ccache ppc64le-objdump: No such file or directory
Fix it by not quoting the string when we expand it, allowing the shell
to do the right thing for us.
Fixes: a71aa05e1416 ("powerpc: Convert relocs_check to a shell script using grep") Fixes: 4ea80652dc75 ("powerpc/64s: Tool to flag direct branches from unrelocated interrupt vectors") Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20191024004730.32135-1-mpe@ellerman.id.au Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Some of our TM (Transactional Memory) tests, list "r1" (the stack
pointer) as a clobbered register.
GCC >= 9 doesn't accept this, and the build breaks:
ptrace-tm-spd-tar.c: In function 'tm_spd_tar':
ptrace-tm-spd-tar.c:31:2: error: listing the stack pointer register 'r1' in a clobber list is deprecated [-Werror=deprecated]
31 | asm __volatile__(
| ^~~
ptrace-tm-spd-tar.c:31:2: note: the value of the stack pointer after an 'asm' statement must be the same as it was before the statement
We do have some fairly large inline asm blocks in these tests, and
some of them do change the value of r1. However they should all return
to C with the value in r1 restored, so I think it's legitimate to say
r1 is not clobbered.
As Segher points out, the r1 clobbers may have been added because of
the use of `or 1,1,1`, however that doesn't actually clobber r1.
Segher also points out that some of these tests do clobber LR, because
they call functions, and that is not listed in the clobbers, so add
that where appropriate.
We should not be putting the chip into reset while interrupts are enabled
and ISR may be running. Fix this by installing a custom devm action and
powering off the device/resetting GPIO line from there. This ensures proper
ordering.
If the hypervisor returned H_PTEG_FULL for H_ENTER hcall, retry a hash page table
insert by removing a random entry from the group.
After some runtime, it is very well possible to find all the 8 hash page table
entry slot in the hpte group used for mapping. Don't fail a bolted entry insert
in that case. With Storage class memory a user can find this error easily since
a namespace enable/disable is equivalent to memory add/remove.
This results in failures as reported below:
$ ndctl create-namespace -r region1 -t pmem -m devdax -a 65536 -s 100M
libndctl: ndctl_dax_enable: dax1.3: failed to enable
Error: namespace1.2: failed to enable
failed to create namespace: No such device or address
In kernel log we find the details as below:
Unable to create mapping for hot added memory 0xc000042006000000..0xc00004200d000000: -1
dax_pmem: probe of dax1.3 failed with error -14
This indicates that we failed to create a bolted hash table entry for direct-map
address backing the namespace.
We also observe failures such that not all namespaces will be enabled with
ndctl enable-namespace all command.
Although if the debugfs initialization fails, we will delete the debugfs
folder of hisi_sas, but we did not consider the scenario where debugfs was
successfully initialized, but the probe failed for other reasons. We found
out that hisi_sas folder is still remain after the probe failed.
When probe fail, we should delete debugfs folder to avoid the above issue.
Link: https://lore.kernel.org/r/1571926105-74636-18-git-send-email-john.garry@huawei.com Signed-off-by: Luo Jiaxing <luojiaxing@huawei.com> Signed-off-by: John Garry <john.garry@huawei.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
For IOs from upper layer, preemption may be disabled as it may be called by
function __blk_mq_delay_run_hw_queue which will call get_cpu() (it disables
preemption). So if flags HISI_SAS_REJECT_CMD_BIT is set in function
hisi_sas_task_exec(), it may disable preempt twice after down() and up()
which will cause following call trace:
When operating in private loop mode, PLOGI exchanges are racing and the
driver tries to abort it's PLOGI. But the PLOGI abort ends up terminating
the login with the other end causing the other end to abort its PLOGI as
well. Discovery never fully completes.
Fix by disabling the PLOGI abort when private loop and letting the state
machine play out.
Link: https://lore.kernel.org/r/20191018211832.7917-5-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
In lpfc_abort_handler, the lock acquire order is hbalock (irqsave),
buf_lock (irq) and ring_lock (irq). The issue is that in two places the
locks are released out of order - the buf_lock and the hbalock - resulting
in the cpu preemption/lock flags getting restored out of order and
deadlocking the cpu.
Fix the unlock order by fully releasing the hbalocks as well.
CC: Zhangguanghui <zhang.guanghui@h3c.com> Link: https://lore.kernel.org/r/20191018211832.7917-7-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
For CHAP [RFC1994], in the first step, the initiator MUST send:
CHAP_A=<A1,A2...>
Where A1,A2... are proposed algorithms, in order of preference.
...
For the Algorithm, as stated in [RFC1994], one value is required to
be implemented:
5 (CHAP with MD5)
LIO currently checks for this value by only comparing a single byte in
the tokenized Algorithm string, which means that any value starting with
a '5' (e.g. "55") is interpreted as "CHAP with MD5". Fix this by
comparing the entire tokenized string.
Reviewed-by: Lee Duncan <lduncan@suse.com> Reviewed-by: Mike Christie <mchristi@redhat.com> Signed-off-by: David Disseldorp <ddiss@suse.de> Link: https://lore.kernel.org/r/20190912095547.22427-2-ddiss@suse.de Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Whenever we reset the channel, we need to clear desc_pendingcount
along with desc_submitcount. Otherwise when a new transaction is
submitted, the irq coalesce level could be programmed to an incorrect
value in the axidma case.
This behavior can be observed when terminating pending transactions
with xilinx_dma_terminate_all() and then submitting new transactions
without releasing and requesting the channel.
Page tables that reside in physical memory beyond the 4 GiB boundary are
currently not working properly. The reason is that when the physical
address for page directory entries is read, it gets truncated at 32 bits
and can cause crashes when passing that address to the DMA API.
Fix this by first casting the PDE value to a dma_addr_t and then using
the page frame number mask for the SMMU instance to mask out the invalid
bits, which are typically used for mapping attributes, etc.
We have added polled mode to the normal input devices with the intent of
retiring input_polled_dev. This converts peaq-wmi driver to use the
polling mode of standard input devices and removes dependency on
INPUT_POLLDEV.
Because the new polling coded does not allow peeking inside the poller
structure to get the poll interval, we change the "debounce" process to
operate on the time basis, instead of counting events.
We also fix error handling during initialization, as previously we leaked
input device structure when we failed to register it.
Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com> Reviewed-by: Hans de Goede <hdegoede@redhat.com> Tested-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Fix warning for:
isst-config.c: In function ‘set_cpu_online_offline’:
isst-config.c:221:3: warning: ignoring return value of ‘write’,
declared with attribute warn_unused_result [-Wunused-result]
write(fd, "1\n", 2);
A validation check to prevent out of bounds read/write inside
functions papr_scm_meta_{get,set}() is off-by-one that prevent reads
and writes to the last byte of the label area.
This bug manifests as a failure to probe a dimm when libnvdimm is
unable to read the entire config-area as advertised by
ND_CMD_GET_CONFIG_SIZE. This usually happens when there are large
number of namespaces created in the region backed by the dimm and the
label-index spans max possible config-area. An error of the form below
usually reported in the kernel logs:
[ 255.293912] nvdimm: probe of nmem0 failed with error -22
The patch fixes these validation checks there by letting libnvdimm
access the entire config-area.
Fixes: 53e80bd042773('powerpc/nvdimm: Add support for multibyte read/write for metadata') Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com> Reviewed-by: Aneesh Kumar K.V <aneesh.kumar@linux.ibm.com> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20190927062002.3169-1-vaibhav@linux.ibm.com Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
generic/018 reports an inconsistent status of atime, the
testcase is as below:
- open file with O_SYNC
- write file to construct fraged space
- calc md5 of file
- record {a,c,m}time
- defrag file --- do nothing
- umount & mount
- check {a,c,m}time
The root cause is, as f2fs enables lazytime by default, atime
update will dirty vfs inode, rather than dirtying f2fs inode (by set
with FI_DIRTY_INODE), so later f2fs_write_inode() called from VFS will
fail to update inode page due to our skip:
f2fs_write_inode()
if (is_inode_flag_set(inode, FI_DIRTY_INODE))
return 0;
So eventually, after evict(), we lose last atime for ever.
To fix this issue, we need to check whether {a,c,m,cr}time is
consistent in between inode cache and inode page, and only skip
f2fs_update_inode() if f2fs inode is not dirty and time is
consistent as well.
Across suspend and resume, we are seeing error messages like the following:
atmel_mxt_ts i2c-PRP0001:00: __mxt_read_reg: i2c transfer failed (-121)
atmel_mxt_ts i2c-PRP0001:00: Failed to read T44 and T5 (-121)
This occurs because the driver leaves its IRQ enabled. Upon resume, there
is an IRQ pending, but the interrupt is serviced before both the driver and
the underlying I2C bus have been resumed. This causes EREMOTEIO errors.
Disable the IRQ in suspend, and re-enable it on resume. If there are cases
where the driver enters suspend with interrupts disabled, that's a bug we
should fix separately.
After study, it was determined there was a double free of a CT iocb during
execution of lpfc_offline_prep and lpfc_offline. The prep routine issued
an abort for some CT iocbs, but the aborts did not complete fast enough for
a subsequent routine that waits for completion. Thus the driver proceeded
to lpfc_offline, which releases any pending iocbs. Unfortunately, the
completions for the aborts were then received which re-released the ct
iocbs.
Turns out the issue for why the aborts didn't complete fast enough was not
their time on the wire/in the adapter. It was the lpfc_work_done routine,
which requires the adapter state to be UP before it calls
lpfc_sli_handle_slow_ring_event() to process the completions. The issue is
the prep routine takes the link down as part of it's processing.
To fix, the following was performed:
- Prevent the offline routine from releasing iocbs that have had aborts
issued on them. Defer to the abort completions. Also means the driver
fully waits for the completions. Given this change, the recognition of
"driver-generated" status which then releases the iocb is no longer
valid. As such, the change made in the commit 296012285c90 is reverted.
As recognition of "driver-generated" status is no longer valid, this
patch reverts the changes made in
commit 296012285c90 ("scsi: lpfc: Fix leak of ELS completions on adapter reset")
- Modify lpfc_work_done to allow slow path completions so that the abort
completions aren't ignored.
- Updated the fdmi path to recognize a CT request that fails due to the
port being unusable. This stops FDMI retries. FDMI will be restarted on
next link up.
Link: https://lore.kernel.org/r/20190922035906.10977-14-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
On some of i.MX SoCs like i.MX8QXP, there is ONLY one IRQ for each
GPIO bank, so it is better to check the IRQ count before getting
second IRQ to avoid below error message during probe:
[ 1.070908] gpio-mxc 5d080000.gpio: IRQ index 1 not found
[ 1.077420] gpio-mxc 5d090000.gpio: IRQ index 1 not found
[ 1.083766] gpio-mxc 5d0a0000.gpio: IRQ index 1 not found
[ 1.090122] gpio-mxc 5d0b0000.gpio: IRQ index 1 not found
[ 1.096470] gpio-mxc 5d0c0000.gpio: IRQ index 1 not found
[ 1.102804] gpio-mxc 5d0d0000.gpio: IRQ index 1 not found
[ 1.109144] gpio-mxc 5d0e0000.gpio: IRQ index 1 not found
[ 1.115475] gpio-mxc 5d0f0000.gpio: IRQ index 1 not found
If any faulty application issues an NVMe Encapsulated commands to HBA which
doesn't support NVMe protocol then driver should return the command as
invalid with the following message.
"HBA doesn't support NVMe. Rejecting NVMe Encapsulated request."
Otherwise below page fault kernel panic will be observed while building the
PRPs as there is no PRP pools allocated for the HBA which doesn't support
NVMe drives.
Symptoms were seen of the driver not having valid data for mailbox
commands. After debugging, the following sequence was found:
The driver maintains a port-wide pointer of the mailbox command that is
currently in execution. Once finished, the port-wide pointer is cleared
(done in lpfc_sli4_mq_release()). The next mailbox command issued will set
the next pointer and so on.
The mailbox response data is only copied if there is a valid port-wide
pointer.
In the failing case, it was seen that a new mailbox command was being
attempted in parallel with the completion. The parallel path was seeing
the mailbox no long in use (flag check under lock) and thus set the port
pointer. The completion path had cleared the active flag under lock, but
had not touched the port pointer. The port pointer is cleared after the
lock is released. In this case, the completion path cleared the just-set
value by the parallel path.
Fix by making the calls that clear mbox state/port pointer while under
lock. Also slightly cleaned up the error path.
Link: https://lore.kernel.org/r/20190922035906.10977-8-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
When user issues diag register command from application with required size,
and if driver unable to allocate the memory, then it will fail the register
command. While failing the register command, driver is not currently
clearing MPT3_CMD_PENDING bit in ctl_cmds.status variable which was set
before trying to allocate the memory. As this bit is set, subsequent
register command will be failed with BUSY status even when user wants to
register the trace buffer will less memory.
Clear MPT3_CMD_PENDING bit in ctl_cmds.status before returning the diag
register command with no memory status.
An issue was seen discovering all SCSI Luns when a target device undergoes
link bounce.
The driver currently does not qualify the FC4 support on the target.
Therefore it will send a SCSI PRLI and an NVMe PRLI. The expectation is
that the target will reject the PRLI if it is not supported. If a PRLI
times out, the driver will retry. The driver will not proceed with the
device until both SCSI and NVMe PRLIs are resolved. In the failure case,
the device is FCP only and does not respond to the NVMe PRLI, thus
initiating the wait/retry loop in the driver. During that time, a RSCN is
received (device bounced) causing the driver to issue a GID_FT. The GID_FT
response comes back before the PRLI mess is resolved and it prematurely
cancels the PRLI retry logic and leaves the device in a STE_PRLI_ISSUE
state. Discovery with the target never completes or resets.
Fix by resetting the node state back to STE_NPR_NODE when GID_FT completes,
thereby restarting the discovery process for the node.
Link: https://lore.kernel.org/r/20190922035906.10977-10-jsmart2021@gmail.com Signed-off-by: Dick Kennedy <dick.kennedy@broadcom.com> Signed-off-by: James Smart <jsmart2021@gmail.com> Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com> Signed-off-by: Sasha Levin <sashal@kernel.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
We need to reset the NIC after setting the bits to enable power
gating and that cannot be done too late in the flow otherwise it
cleans other registers and things that were already configured,
causing initialization to fail.
In order to fix this, move the function to the common code in trans.c
so it can be called directly from there at an earlier point, just
after the reset we already do during initialization.
Fixes: 9a47cb988338 ("iwlwifi: pcie: add workaround for power gating in integrated 22000")
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=205719 Cc: stable@ver.kernel.org # 5.4+ Reported-by: Anders Kaseorg <andersk@mit.edu> Signed-off-by: Luca Coelho <luciano.coelho@intel.com> Signed-off-by: Kalle Valo <kvalo@codeaurora.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
where we can deadlock during device shutdown. The problem occurs if
the recv_work's nbd_config_put occurs after nbd_start_device_ioctl has
returned and the userspace app has droppped its reference via closing
the device and running nbd_release. The recv_work nbd_config_put call
would then drop the refcount to zero and try to destroy the config which
would try to do destroy_workqueue from the recv work.
This patch just has nbd_start_device_ioctl do a flush_workqueue when it
wakes so we know after the ioctl returns running works have exited. This
also fixes a possible race where we could try to reuse the device while
old recv_works are still running.
Cc: stable@vger.kernel.org Fixes: e9e006f5fcf2 ("nbd: fix max number of supported devs") Signed-off-by: Mike Christie <mchristi@redhat.com> Signed-off-by: Jens Axboe <axboe@kernel.dk> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Command queuing has been reported broken on some systems based on Intel
GLK. A separate patch disables command queuing in some cases.
This patch adds a quirk for broken command queuing, which enables users
with problems to disable command queuing using sdhci module parameters for
quirks.
Fixes: 8ee82bda230f ("mmc: sdhci-pci: Add CQHCI support for Intel GLK") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20191217095349.14592-2-adrian.hunter@intel.com Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Command queuing has been reported broken on some Lenovo systems based on
Intel GLK. This is likely a BIOS issue, so disable command queuing for
Intel GLK if the BIOS vendor string is "LENOVO".
Fixes: 8ee82bda230f ("mmc: sdhci-pci: Add CQHCI support for Intel GLK") Signed-off-by: Adrian Hunter <adrian.hunter@intel.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20191217095349.14592-1-adrian.hunter@intel.com Signed-off-by: Ulf Hansson <ulf.hansson@linaro.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Two previous patches introduced below quirks for P2020 platforms.
- SDHCI_QUIRK_RESET_AFTER_REQUEST
- SDHCI_QUIRK_BROKEN_TIMEOUT_VAL
The patches made a mistake to add them in quirks2 of sdhci_host
structure, while they were defined for quirks.
host->quirks2 |= SDHCI_QUIRK_RESET_AFTER_REQUEST;
host->quirks2 |= SDHCI_QUIRK_BROKEN_TIMEOUT_VAL;
This patch is to fix them.
host->quirks |= SDHCI_QUIRK_RESET_AFTER_REQUEST;
host->quirks |= SDHCI_QUIRK_BROKEN_TIMEOUT_VAL;
Tuning support in DDR50 speed mode was added in SD Specifications Part1
Physical Layer Specification v3.01. Its not possible to distinguish
between v3.00 and v3.01 from the SCR and that is why since
commit 4324f6de6d2e ("mmc: core: enable CMD19 tuning for DDR50 mode")
tuning failures are ignored in DDR50 speed mode.
Cards compatible with v3.00 don't respond to CMD19 in DDR50 and this
error gets printed during enumeration and also if retune is triggered at
any time during operation. Update the printk level to pr_debug so that
these errors don't lead to false error reports.
First, the fix seems to be plain wrong, since the erratum suggests
waiting 5ms before setting setting SYSCTL[RSTD], but this msleep()
happens after the call of sdhci_reset() which is where that bit gets
set (if SDHCI_RESET_DATA is in mask).
Second, walking the whole device tree to figure out if some node has a
"fsl,p2020-esdhc" compatible string is hugely expensive - about 70 to
100 us on our mpc8309 board. Walking the device tree is done under a
raw_spin_lock, so this is obviously really bad on an -rt system, and a
waste of time on all.
In fact, since esdhc_reset() seems to get called around 100 times per
second, that mpc8309 now spends 0.8% of its time determining that
it is not a p2020. Whether those 100 calls/s are normal or due to some
other bug or misconfiguration, regularly hitting a 100 us
non-preemptible window is unacceptable.
The DDR_CONFIG register offset got updated after a specific
minor version of sdcc V4. This offset change has not been properly
taken care of while updating register changes for sdcc V5.
Correcting proper offset for this register.
Also updating this register value to reflect the recommended RCLK
delay.
If an ocxl device is unbound through sysfs at the same time its AFU is
being opened by a user process, the open code may dereference freed
stuctures, which can lead to kernel oops messages. You'd have to hit a
tiny time window, but it's possible. It's fairly easy to test by
making the time window bigger artificially.
Fix it with a combination of 2 changes:
- when an AFU device is found in the IDR by looking for the device
minor number, we should hold a reference on the device until after
the context is allocated. A reference on the AFU structure is kept
when the context is allocated, so we can release the reference on
the device after the context allocation.
- with the fix above, there's still another even tinier window,
between the time the AFU device is found in the IDR and the
reference on the device is taken. We can fix this one by removing
the IDR entry earlier, when the device setup is removed, instead
of waiting for the 'release' device callback. With proper locking
around the IDR.
Fixes: 75ca758adbaf ("ocxl: Create a clear delineation between ocxl backend & frontend") Cc: stable@vger.kernel.org # v5.2+ Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com> Reviewed-by: Greg Kurz <groug@kaod.org> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20190624144148.32022-1-fbarrat@linux.ibm.com Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Before commit 0366a1c70b89 ("powerpc/irq: Run softirqs off the top of
the irq stack"), check_stack_overflow() was called by do_IRQ(), before
switching to the irq stack.
In that commit, do_IRQ() was renamed __do_irq(), and is now executing
on the irq stack, so check_stack_overflow() has just become almost
useless.
Move check_stack_overflow() call in do_IRQ() to do the check while
still on the current stack.
Fixes: 0366a1c70b89 ("powerpc/irq: Run softirqs off the top of the irq stack") Cc: stable@vger.kernel.org Signed-off-by: Christophe Leroy <christophe.leroy@c-s.fr> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/e033aa8116ab12b7ca9a9c75189ad0741e3b9b5f.1575872340.git.christophe.leroy@c-s.fr Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
With commit 247f2f6f3c70 ("sched/core: Don't schedule threads on
pre-empted vCPUs"), the scheduler avoids preempted vCPUs to schedule
tasks on wakeup. This leads to wrong choice of CPU, which in-turn
leads to larger wakeup latencies. Eventually, it leads to performance
regression in latency sensitive benchmarks like soltp, schbench etc.
On Powerpc, vcpu_is_preempted() only looks at yield_count. If the
yield_count is odd, the vCPU is assumed to be preempted. However
yield_count is increased whenever the LPAR enters CEDE state (idle).
So any CPU that has entered CEDE state is assumed to be preempted.
Even if vCPU of dedicated LPAR is preempted/donated, it should have
right of first-use since they are supposed to own the vCPU.
On a Power9 System with 32 cores:
# lscpu
Architecture: ppc64le
Byte Order: Little Endian
CPU(s): 128
On-line CPU(s) list: 0-127
Thread(s) per core: 8
Core(s) per socket: 1
Socket(s): 16
NUMA node(s): 2
Model: 2.2 (pvr 004e 0202)
Model name: POWER9 (architected), altivec supported
Hypervisor vendor: pHyp
Virtualization type: para
L1d cache: 32K
L1i cache: 32K
L2 cache: 512K
L3 cache: 10240K
NUMA node0 CPU(s): 0-63
NUMA node1 CPU(s): 64-127
The function mce_severity_amd_smca() requires m->bank to be initialized
for correct operation. Fix the one case, where mce_severity() is called
without doing so.
Fixes: 6bda529ec42e ("x86/mce: Grade uncorrected errors for SMCA-enabled systems") Fixes: d28af26faa0b ("x86/MCE: Initialize mce.bank in the case of a fatal error in mce_no_way_out()") Signed-off-by: Jan H. Schönherr <jschoenh@amazon.de> Signed-off-by: Borislav Petkov <bp@suse.de> Reviewed-by: Tony Luck <tony.luck@intel.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@kernel.org> Cc: linux-edac <linux-edac@vger.kernel.org> Cc: <stable@vger.kernel.org> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: x86-ml <x86@kernel.org> Cc: Yazen Ghannam <Yazen.Ghannam@amd.com> Link: https://lkml.kernel.org/r/20191210000733.17979-4-jschoenh@amazon.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Each logical CPU in Scalable MCA systems controls a unique set of MCA
banks in the system. These banks are not shared between CPUs. The bank
types and ordering will be the same across CPUs on currently available
systems.
However, some CPUs may see a bank as Reserved/Read-as-Zero (RAZ) while
other CPUs do not. In this case, the bank seen as Reserved on one CPU is
assumed to be the same type as the bank seen as a known type on another
CPU.
In general, this occurs when the hardware represented by the MCA bank
is disabled, e.g. disabled memory controllers on certain models, etc.
The MCA bank is disabled in the hardware, so there is no possibility of
getting an MCA/MCE from it even if it is assumed to have a known type.
For example:
Full system:
Bank | Type seen on CPU0 | Type seen on CPU1
------------------------------------------------
0 | LS | LS
1 | UMC | UMC
2 | CS | CS
System with hardware disabled:
Bank | Type seen on CPU0 | Type seen on CPU1
------------------------------------------------
0 | LS | LS
1 | UMC | RAZ
2 | CS | CS
For this reason, there is a single, global struct smca_banks[] that is
initialized at boot time. This array is initialized on each CPU as it
comes online. However, the array will not be updated if an entry already
exists.
This works as expected when the first CPU (usually CPU0) has all
possible MCA banks enabled. But if the first CPU has a subset, then it
will save a "Reserved" type in smca_banks[]. Successive CPUs will then
not be able to update smca_banks[] even if they encounter a known bank
type.
This may result in unexpected behavior. Depending on the system
configuration, a user may observe issues enumerating the MCA
thresholding sysfs interface. The issues may be as trivial as sysfs
entries not being available, or as severe as system hangs.
For example:
Bank | Type seen on CPU0 | Type seen on CPU1
------------------------------------------------
0 | LS | LS
1 | RAZ | UMC
2 | CS | CS
Extend the smca_banks[] entry check to return if the entry is a
non-reserved type. Otherwise, continue so that CPUs that encounter a
known bank type can update smca_banks[].
... because interrupts are disabled that early and sending IPIs can
deadlock:
BUG: sleeping function called from invalid context at kernel/sched/completion.c:99
in_atomic(): 1, irqs_disabled(): 1, non_block: 0, pid: 0, name: swapper/1
no locks held by swapper/1/0.
irq event stamp: 0
hardirqs last enabled at (0): [<0000000000000000>] 0x0
hardirqs last disabled at (0): [<ffffffff8106dda9>] copy_process+0x8b9/0x1ca0
softirqs last enabled at (0): [<ffffffff8106dda9>] copy_process+0x8b9/0x1ca0
softirqs last disabled at (0): [<0000000000000000>] 0x0
Preemption disabled at:
[<ffffffff8104703b>] start_secondary+0x3b/0x190
CPU: 1 PID: 0 Comm: swapper/1 Not tainted 5.5.0-rc2+ #1
Hardware name: GIGABYTE MZ01-CE1-00/MZ01-CE1-00, BIOS F02 08/29/2018
Call Trace:
dump_stack
___might_sleep.cold.92
wait_for_completion
? generic_exec_single
rdmsr_safe_on_cpu
? wrmsr_on_cpus
mce_amd_feature_init
mcheck_cpu_init
identify_cpu
identify_secondary_cpu
smp_store_cpu_info
start_secondary
secondary_startup_64
The function smca_configure() is called only on the current CPU anyway,
therefore replace rdmsr_safe_on_cpu() with atomic rdmsr_safe() and avoid
the IPI.
Commit 4b927b94d5df ("KVM: arm/arm64: vgic: Introduce find_reg_by_id()")
introduced 'find_reg_by_id()', which looks up a system register only if
the 'id' index parameter identifies a valid system register. As part of
the patch, existing callers of 'find_reg()' were ported over to the new
interface, but this breaks 'index_to_sys_reg_desc()' in the case that the
initial lookup in the vCPU target table fails because we will then call
into 'find_reg()' for the system register table with an uninitialised
'param' as the key to the lookup.
GCC 10 is bright enough to spot this (amongst a tonne of false positives,
but hey!):
| arch/arm64/kvm/sys_regs.c: In function ‘index_to_sys_reg_desc.part.0.isra’:
| arch/arm64/kvm/sys_regs.c:983:33: warning: ‘params.Op2’ may be used uninitialized in this function [-Wmaybe-uninitialized]
| 983 | (u32)(x)->CRn, (u32)(x)->CRm, (u32)(x)->Op2);
| [...]
Revert the hunk of 4b927b94d5df which breaks 'index_to_sys_reg_desc()' so
that the old behaviour of checking the index upfront is restored.
A device mapping is normally always mapped at Stage-2, since there
is very little gain in having it faulted in.
Nonetheless, it is possible to end-up in a situation where the device
mapping has been removed from Stage-2 (userspace munmaped the VFIO
region, and the MMU notifier did its job), but present in a userspace
mapping (userpace has mapped it back at the same address). In such
a situation, the device mapping will be demand-paged as the guest
performs memory accesses.
This requires to be careful when dealing with mapping size, cache
management, and to handle potential execution of a device mapping.
Reported-by: Alexandru Elisei <alexandru.elisei@arm.com> Signed-off-by: Marc Zyngier <maz@kernel.org> Tested-by: Alexandru Elisei <alexandru.elisei@arm.com> Reviewed-by: James Morse <james.morse@arm.com> Cc: stable@vger.kernel.org Link: https://lore.kernel.org/r/20191211165651.7889-2-maz@kernel.org Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
The host reports support for the synthetic feature X86_FEATURE_SSBD
when any of the three following hardware features are set:
CPUID.(EAX=7,ECX=0):EDX.SSBD[bit 31]
CPUID.80000008H:EBX.AMD_SSBD[bit 24]
CPUID.80000008H:EBX.VIRT_SSBD[bit 25]
Either of the first two hardware features implies the existence of the
IA32_SPEC_CTRL MSR, but CPUID.80000008H:EBX.VIRT_SSBD[bit 25] does
not. Therefore, CPUID.80000008H:EBX.AMD_SSBD[bit 24] should only be
set in the guest if CPUID.(EAX=7,ECX=0):EDX.SSBD[bit 31] or
CPUID.80000008H:EBX.AMD_SSBD[bit 24] is set on the host.
Fixes: 4c6903a0f9d76 ("KVM: x86: fix reporting of AMD speculation bug CPUID leaf") Signed-off-by: Jim Mattson <jmattson@google.com> Reviewed-by: Jacob Xu <jacobhxu@google.com> Reviewed-by: Peter Shier <pshier@google.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: stable@vger.kernel.org Reported-by: Eric Biggers <ebiggers@kernel.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
The host reports support for the synthetic feature X86_FEATURE_SSBD
when any of the three following hardware features are set:
CPUID.(EAX=7,ECX=0):EDX.SSBD[bit 31]
CPUID.80000008H:EBX.AMD_SSBD[bit 24]
CPUID.80000008H:EBX.VIRT_SSBD[bit 25]
Either of the first two hardware features implies the existence of the
IA32_SPEC_CTRL MSR, but CPUID.80000008H:EBX.VIRT_SSBD[bit 25] does
not. Therefore, CPUID.(EAX=7,ECX=0):EDX.SSBD[bit 31] should only be
set in the guest if CPUID.(EAX=7,ECX=0):EDX.SSBD[bit 31] or
CPUID.80000008H:EBX.AMD_SSBD[bit 24] is set on the host.
Fixes: 0c54914d0c52a ("KVM: x86: use Intel speculation bugs and features as derived in generic x86 code") Signed-off-by: Jim Mattson <jmattson@google.com> Reviewed-by: Jacob Xu <jacobhxu@google.com> Reviewed-by: Peter Shier <pshier@google.com> Cc: Paolo Bonzini <pbonzini@redhat.com> Cc: stable@vger.kernel.org Reported-by: Eric Biggers <ebiggers@kernel.org> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
VCPU_CR is the offset of arch.regs.ccr in kvm_vcpu.
arch/powerpc/include/asm/kvm_host.h defines arch.regs as a struct
pt_regs, and arch/powerpc/include/asm/ptrace.h defines the ccr field
of pt_regs as "unsigned long ccr". Since unsigned long is 64 bits, a
64-bit load needs to be used to load it, unless an endianness specific
correction offset is added to access the desired subpart. In this
case there is no reason to _not_ use a 64 bit load though.
Fixes: 6c85b7bc637b ("powerpc/kvm: Use UV_RETURN ucall to return to ultravisor") Cc: stable@vger.kernel.org # v5.4+ Signed-off-by: Marcus Comstedt <marcus@mc.pp.se> Signed-off-by: Michael Ellerman <mpe@ellerman.id.au> Link: https://lore.kernel.org/r/20191215094900.46740-1-marcus@mc.pp.se Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
When over-budget IOs are force-issued through root cgroup,
iocg_kick_delay() adjusts the async delay accordingly but doesn't
actually schedule async throttle for the issuing task. This bug is
pretty well masked because sooner or later the offending threads are
gonna get directly throttled on regular IOs or have async delay
scheduled by mem_cgroup_throttle_swaprate().
However, it can affect control quality on filesystem metadata heavy
operations. Let's fix it by invoking blkcg_schedule_throttle() when
iocg_kick_delay() says async delay is needed.
Instead of setting s_want_extra_size and then making sure that it is a
valid value afterwards, validate the field before we set it. This
avoids races and other problems when remounting the file system.
ext4_check_dir_entry() currently does not catch a case when a directory
entry ends so close to the block end that the header of the next
directory entry would not fit in the remaining space. This can lead to
directory iteration code trying to access address beyond end of current
buffer head leading to oops.
CC: stable@vger.kernel.org Signed-off-by: Jan Kara <jack@suse.cz> Link: https://lore.kernel.org/r/20191202170213.4761-3-jack@suse.cz Signed-off-by: Theodore Ts'o <tytso@mit.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Function ext4_empty_dir() doesn't correctly handle directories with
holes and crashes on bh->b_data dereference when bh is NULL. Reorganize
the loop to use 'offset' variable all the times instead of comparing
pointers to current direntry with bh->b_data pointer. Also add more
strict checking of '.' and '..' directory entries to avoid entering loop
in possibly invalid state on corrupted filesystems.
The usage of readl_poll_timeout is wrong, the 3rd parameter(cond)
should be "val & LOCK_STATUS" not "val & LOCK_TIMEOUT_US",
It is not check whether the pll locked, LOCK_STATUS reflects the mask,
not LOCK_TIMEOUT_US.
Fixes: 8646d4dcc7fb ("clk: imx: Add PLLs driver for imx8mm soc") Cc: <stable@vger.kernel.org> Reviewed-by: Abel Vesa <abel.vesa@nxp.com> Signed-off-by: Peng Fan <peng.fan@nxp.com> Signed-off-by: Shawn Guo <shawnguo@kernel.org> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
There is a lock to divider in the composite driver, but that's not
enough. lock to gate/mux are also needed to provide exclusive access
to the register.
Commit 39ce8150a079 ("pinctrl: baytrail: Serialize all register access")
added a spinlock around all register accesses because:
"There is a hardware issue in Intel Baytrail where concurrent GPIO register
access might result reads of 0xffffffff and writes might get dropped
completely."
Testing has shown that this does not catch all cases, there are still
2 problems remaining
1) The original fix uses a spinlock per byt_gpio device / struct,
additional testing has shown that this is not sufficient concurent
accesses to 2 different GPIO banks also suffer from the same problem.
This commit fixes this by moving to a single global lock.
2) The original fix did not add a lock around the register accesses in
the suspend/resume handling.
Since pinctrl-baytrail.c is using normal suspend/resume handlers,
interrupts are still enabled during suspend/resume handling. Nothing
should be using the GPIOs when they are being taken down, _but_ the
GPIOs themselves may still cause interrupts, which are likely to
use (read) the triggering GPIO. So we need to protect against
concurrent GPIO register accesses in the suspend/resume handlers too.
This commit fixes this by adding the missing spin_lock / unlock calls.
The 2 fixes together fix the Acer Switch 10 SW5-012 getting completely
confused after a suspend resume. The DSDT for this device has a bug
in its _LID method which reprograms the home and power button trigger-
flags requesting both high and low _level_ interrupts so the IRQs for
these 2 GPIOs continuously fire. This combined with the saving of
registers during suspend, triggers concurrent GPIO register accesses
resulting in saving 0xffffffff as pconf0 value during suspend and then
when restoring this on resume the pinmux settings get all messed up,
resulting in various I2C busses being stuck, the wifi no longer working
and often the tablet simply not coming out of suspend at all.
Cc: stable@vger.kernel.org Fixes: 39ce8150a079 ("pinctrl: baytrail: Serialize all register access") Signed-off-by: Hans de Goede <hdegoede@redhat.com> Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
A break interrupt will be generated if the RX line was pulled low, which
means some abnomal behaviors occurred of the UART. In this case, we still
need to clear this break interrupt status, otherwise it will cause irq
storm to crash the whole system.
The "auto-attach" handler function `gsc_hpdi_auto_attach()` calls
`dma_alloc_coherent()` in a loop to allocate some DMA data buffers, and
also calls it to allocate a buffer for a DMA descriptor chain. However,
it does not check the return value of any of these calls. Change
`gsc_hpdi_auto_attach()` to return `-ENOMEM` if any of these
`dma_alloc_coherent()` calls fail. This will result in the comedi core
calling the "detach" handler `gsc_hpdi_detach()` as part of the
clean-up, which will call `gsc_hpdi_free_dma()` to free any allocated
DMA coherent memory buffers.
At least on the HP Envy x360 15-cp0xxx model the WMI interface
for HPWMI_FEATURE2_QUERY requires an outsize of at least 128 bytes,
otherwise it fails with an error code 5 (HPWMI_RET_INVALID_PARAMETERS):
Dec 06 00:59:38 kernel: hp_wmi: query 0xd returned error 0x5
We do not care about the contents of the buffer, we just want to know
if the HPWMI_FEATURE2_QUERY command is supported.
This commits bumps the buffer size, fixing the error.
Fixes: 8a1513b4932 ("hp-wmi: limit hotkey enable") Cc: stable@vger.kernel.org BugLink: https://bugzilla.redhat.com/show_bug.cgi?id=1520703 Signed-off-by: Hans de Goede <hdegoede@redhat.com> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Seth Forshee <seth.forshee@canonical.com>
Commit 6cac7866c2741 ("intel_th: msu: Add a sysfs attribute to trigger
window switch") adds a NULL pointer dereference in the case when there are
no windows allocated:
Commit aac8da65174a ("intel_th: msu: Start handling IRQs") implicitly
relies on the use of devm_request_irq() to subsequently free the irqs on
device removal, but in case of the pci_free_irq_vectors() API, the
handlers need to be freed before it is called. Therefore, at the moment
the driver's remove path trips a BUG_ON(irq_has_action()):