| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
There is a extraneous space after a newline in a dev_err message.
Remove it.
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The master ooo cannot be completely closed when the
accelerator core reports memory error. Therefore, the driver
needs to inject the qm error to close the master ooo. Currently,
the qm error is injected after stopping queue, memory may be
released immediately after stopping queue, causing the device to
access the released memory. Therefore, error is injected to close master
ooo before stopping queue to ensure that the device does not access
the released memory.
Fixes: 6c6dd5802c2d ("crypto: hisilicon/qm - add controller reset interface")
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The timeout threshold of the hpre cluster is 16ms. When the CPU
and device share virtual address, page fault processing time may
exceed the threshold.
In the current test, there is a high probability that the
cluster times out. However, the cluster is waiting for the
completion of memory access, which is not an error, the device
does not need to be reset. If an error occurs in the cluster,
qm also reports the error. Therefore, the cluster timeout
error of hpre can be masked.
Fixes: d90fab0deb8e ("crypto: hisilicon/qm - get error type from hardware registers")
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Before the device is enabled again, the device may still
store the previously processed data. If an error occurs in
the previous task, the device may fail to be enabled again.
Therefore, before enabling device, reset the device to restore
the initial state.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Header files is included Order-ref: standard library headers,
OS library headers, and project-specific headers. This patch
modifies the order of header files according to suggestions.
In addition, use %u to print unsigned int variables to prevent
overflow.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Apply for a lock before the qp send operation to ensure no
resource race in multi-concurrency situations.
This modification has almost no impact on performance.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
If an error occurs in the process after the SGL is mapped
successfully, it need to unmap the SGL.
Otherwise, memory problems may occur.
Signed-off-by: Yang Shen <[email protected]>
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The req_lock is currently implemented as a rw_lock, but there are no
instances where read_lock() is called. This means that the lock is
effectively only used by writers, making it functionally equivalent to
a simple spinlock.
As stated in Documentation/locking/spinlocks.rst:
"Reader-writer locks require more atomic memory operations than simple
spinlocks. Unless the reader critical section is long, you are better
off just using spinlocks."
Since the rw_lock in this case incurs additional atomic memory
operations without any benefit from reader-writer locking, it is more
efficient to replace it with a spinlock. This patch implements that
replacement to optimize the driver's performance.
Signed-off-by: Kuan-Wei Chiu <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Currently, the reg is queried based on the fixed address offset
array. When the number of accelerator cores changes, the system
can not flexibly respond to the change.
Therefore, the reg to be queried is calculated based on the
comp or decomp core base address.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
enable and disable
When the vf is enabled, the value of vfs_num must be assigned
after the VF configuration is complete. Otherwise, the device
may be accessed before the virtual configuration is complete,
causing an error.
When the vf is disabled, clear vfs_num and execute
qm_pm_put_sync before hisi_qm_sriov_disable is return.
Otherwise, if qm_clear_vft_config fails, users may access the
device when the PCI virtualization is disabled, resulting in an
error.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The offset of SEC_CORE_ENABLE_BITMAP should be 0 instead of 32,
it cause a kasan shift-out-bounds warning, fix it.
Signed-off-by: Wenkai Lin <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Some information showed by the dump function is invalid. Mask
the unnecessary information from the dump file.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
When the qm uninit command is executed, the err data needs to
be released to prevent memory leakage. The error information
release operation and uacce_remove are integrated in
qm_remove_uacce.
So add the qm_remove_uacce to qm uninit to avoid err memory
leakage.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
space in sq dump
When dumping SQ, only the corresponding ID's SQE needs to be
dumped, and there is no need to apply for the entire SQE
memory. This is because excessive dump operations can lead to
memory resource waste.
Therefor apply for the space corresponding to sqe_id separately
to avoid space waste.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The AIV is one of the SEC resources. When releasing resources,
it need to release the AIV resources at the same time.
Otherwise, memory leakage occurs.
The aiv resource release is added to the sec resource release
function.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
There is a scenario where the file directory is created but the
file memory is not set. In this case, if a user accesses the
file, an error occurs.
So during the creation process of debugfs, memory should be
allocated first before creating the directory. In the release
process, the directory should be deleted first before releasing
the memory to avoid the situation where the memory does not
exist when accessing the directory.
In addition, the directory released by the debugfs is a global
variable. When the debugfs of an accelerator fails to be
initialized, releasing the directory of the global variable
affects the debugfs initialization of other accelerators.
The debugfs root directory released by debugfs init should be a
member of qm, not a global variable.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The cmd type can be extended. Currently, only four types of cmd
can be processed. Therefor, add the default processing branch
to intercept incorrect parameter input.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
creation
There is a scenario where the file directory is created but the
file attribute is not set. In this case, if a user accesses the
file, an error occurs.
So adjust the processing logic in the debugfs creation to
prevent the file from being accessed before the file attributes
such as the index are set.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The input parameter check in acc_get_sgl is redundant. The
caller has been verified once. When the check is performed for
multiple times, the performance deteriorates.
So the redundant parameter verification is deleted, and the
index verification is changed to the module entry function for
verification.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
During the zip probe process, the debugfs failure does not stop
the probe. When debugfs initialization fails, jumping to the
error branch will also release regs, in addition to its own
rollback operation.
As a result, it may be released repeatedly during the regs
uninit process. Therefore, the null check needs to be added to
the regs uninit process.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
When CONFIG_PCI_IOV is disabled, the SRIOV configuration
function is not required. An error occurs if this function is
incorrectly called.
Consistent with other modules, add the condition for
configuring the sriov function of sec_pci_driver.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Add the missing CRYPTO_ALG_ASYNC flag since hisilizon zip driver
works asynchronously.
Cc: Zhou Wang <[email protected]>
Signed-off-by: Barry Song <[email protected]>
Acked-by: Yang Shen <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The function qm_stop_qp_nolock() always return zero, so
function type is changed to void.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The debugfs files 'dev_state' and 'dev_timeout' are added.
Users can query the current queue stop status through these two
files. And set the waiting timeout when the queue is released.
dev_state: if dev_timeout is set, dev_state indicates the status
of stopping the queue. 0 indicates that the queue is stopped
successfully. Other values indicate that the queue stops fail.
If dev_timeout is not set, the value of dev_state is 0;
dev_timeout: if the queue fails to stop, the queue is released
after waiting dev_timeout * 20ms.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Hardware V3 could be able to drain function by sending mailbox
to hardware which will trigger tasks in device to be flushed out.
When the function is reset, the function can be stopped by this way.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Unused parameter of static functions should be removed.
Signed-off-by: Wenkai Lin <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
This patch fixes following cleanup issues:
- The return value of the function is
inconsistent with the actual return type.
- After the pointer type is directly converted
to the `__le64` type, the program may crash
or produce unexpected results.
Signed-off-by: Qi Tao <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Nested macros are integrated into a single macro,
making the code simpler.
Signed-off-by: Qi Tao <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
As the sec DFX function is enhanced, some RAS registers are added
to the original DFX registers to enhance the DFX positioning function.
Signed-off-by: Qi Tao <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Switch to raw_smp_processor_id() to prevent a number of
warnings from kernel debugging. We do not care about
preemption here, as the CPU number is only used as a
poor mans load balancing or device selection. If preemption
happens during an encrypt/decrypt operation a small performance
hit will occur but everything will continue to work, so just
ignore it. This commit is similar to e7a9b05ca4
("crypto: cavium - Fix smp_processor_id() warnings").
[ 7538.874350] BUG: using smp_processor_id() in preemptible [00000000] code: af_alg06/8438
[ 7538.874368] caller is debug_smp_processor_id+0x1c/0x28
[ 7538.874373] CPU: 50 PID: 8438 Comm: af_alg06 Kdump: loaded Not tainted 5.10.0.pc+ #18
[ 7538.874377] Call trace:
[ 7538.874387] dump_backtrace+0x0/0x210
[ 7538.874389] show_stack+0x2c/0x38
[ 7538.874392] dump_stack+0x110/0x164
[ 7538.874394] check_preemption_disabled+0xf4/0x108
[ 7538.874396] debug_smp_processor_id+0x1c/0x28
[ 7538.874406] sec_create_qps+0x24/0xe8 [hisi_sec2]
[ 7538.874408] sec_ctx_base_init+0x20/0x4d8 [hisi_sec2]
[ 7538.874411] sec_aead_ctx_init+0x68/0x180 [hisi_sec2]
[ 7538.874413] sec_aead_sha256_ctx_init+0x28/0x38 [hisi_sec2]
[ 7538.874421] crypto_aead_init_tfm+0x54/0x68
[ 7538.874423] crypto_create_tfm_node+0x6c/0x110
[ 7538.874424] crypto_alloc_tfm_node+0x74/0x288
[ 7538.874426] crypto_alloc_aead+0x40/0x50
[ 7538.874431] aead_bind+0x50/0xd0
[ 7538.874433] alg_bind+0x94/0x148
[ 7538.874439] __sys_bind+0x98/0x118
[ 7538.874441] __arm64_sys_bind+0x28/0x38
[ 7538.874445] do_el0_svc+0x88/0x258
[ 7538.874447] el0_svc+0x1c/0x28
[ 7538.874449] el0_sync_handler+0x8c/0xb8
[ 7538.874452] el0_sync+0x148/0x180
Signed-off-by: Wenkai Lin <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Read the values of some device registers before the device
is reset, these values help analyze the cause of the device exception.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Support get device current state. The value 0 indicates that
the device is busy, and the value 1 indicates that the
device is idle. When the device is in suspended, 1 is returned.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Pull crypto updates from Herbert Xu:
"API:
- Add incremental lskcipher/skcipher processing
Algorithms:
- Remove SHA1 from drbg
- Remove CFB and OFB
Drivers:
- Add comp high perf mode configuration in hisilicon/zip
- Add support for 420xx devices in qat
- Add IAA Compression Accelerator driver"
* tag 'v6.8-p1' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (172 commits)
crypto: iaa - Account for cpu-less numa nodes
crypto: scomp - fix req->dst buffer overflow
crypto: sahara - add support for crypto_engine
crypto: sahara - remove error message for bad aes request size
crypto: sahara - remove unnecessary NULL assignments
crypto: sahara - remove 'active' flag from sahara_aes_reqctx struct
crypto: sahara - use dev_err_probe()
crypto: sahara - use devm_clk_get_enabled()
crypto: sahara - use BIT() macro
crypto: sahara - clean up macro indentation
crypto: sahara - do not resize req->src when doing hash operations
crypto: sahara - fix processing hash requests with req->nbytes < sg->length
crypto: sahara - improve error handling in sahara_sha_process()
crypto: sahara - fix wait_for_completion_timeout() error handling
crypto: sahara - fix ahash reqsize
crypto: sahara - handle zero-length aes requests
crypto: skcipher - remove excess kerneldoc members
crypto: shash - remove excess kerneldoc members
crypto: qat - generate dynamically arbiter mappings
crypto: qat - add support for ring pair level telemetry
...
|
|
commit 23baf831a32c ("mm, treewide: redefine MAX_ORDER sanely") has
changed the definition of MAX_ORDER to be inclusive. This has caused
issues with code that was not yet upstream and depended on the previous
definition.
To draw attention to the altered meaning of the define, rename MAX_ORDER
to MAX_PAGE_ORDER.
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Kirill A. Shutemov <[email protected]>
Cc: Linus Torvalds <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>
|
|
Add the printf of an error message and optimized the handling
process of ret.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Deleted a dbg function because this function has the risk of
address leakage. In addition, this function is only used for
debugging in the early stage and is not required in the future.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Remove the unused CFB/OFB implementation.
Signed-off-by: Herbert Xu <[email protected]>
|
|
Pre-store the valid value of the zip alg support related capability
register in hisi_zip_qm_init(), which will be called by hisi_zip_probe().
It can reduce the number of capability register queries and avoid
obtaining incorrect values in abnormal scenarios, such as reset failed
and the memory space disabled.
Fixes: db700974b69d ("crypto: hisilicon/zip - support zip capability")
Signed-off-by: Zhiqi Song <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Pre-store the valid value of the sec alg support related capability
register in sec_qm_init(), which will be called by probe process.
It can reduce the number of capability register queries and avoid
obtaining incorrect values in abnormal scenarios, such as reset
failed and the memory space disabled.
Fixes: 921715b6b782 ("crypto: hisilicon/sec - get algorithm bitmap from registers")
Signed-off-by: Zhiqi Song <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Pre-store the valid value of hpre alg support related capability
register in hpre_qm_init(), which will be called by hpre_probe().
It can reduce the number of capability register queries and avoid
obtaining incorrect values in abnormal scenarios, such as reset
failed and the memory space disabled.
Fixes: f214d59a0603 ("crypto: hisilicon/hpre - support hpre capability")
Signed-off-by: Zhiqi Song <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
In previous capability register implementation, qm irq related values
were read from capability registers dynamically when needed. But in
abnormal scenario, e.g. the core is timeout and the device needs to
soft reset and reset failed after disabling the MSE, the device can
not be removed normally, causing the following call trace:
| Call trace:
| pci_irq_vector+0xfc/0x140
| hisi_qm_uninit+0x278/0x3b0 [hisi_qm]
| hpre_remove+0x16c/0x1c0 [hisi_hpre]
| pci_device_remove+0x6c/0x264
| device_release_driver_internal+0x1ec/0x3e0
| device_release_driver+0x3c/0x60
| pci_stop_bus_device+0xfc/0x22c
| pci_stop_and_remove_bus_device+0x38/0x70
| pci_iov_remove_virtfn+0x108/0x1c0
| sriov_disable+0x7c/0x1e4
| pci_disable_sriov+0x4c/0x6c
| hisi_qm_sriov_disable+0x90/0x160 [hisi_qm]
| hpre_remove+0x1a8/0x1c0 [hisi_hpre]
| pci_device_remove+0x6c/0x264
| device_release_driver_internal+0x1ec/0x3e0
| driver_detach+0x168/0x2d0
| bus_remove_driver+0xc0/0x230
| driver_unregister+0x58/0xdc
| pci_unregister_driver+0x40/0x220
| hpre_exit+0x34/0x64 [hisi_hpre]
| __arm64_sys_delete_module+0x374/0x620
[...]
| Call trace:
| free_msi_irqs+0x25c/0x300
| pci_disable_msi+0x19c/0x264
| pci_free_irq_vectors+0x4c/0x70
| hisi_qm_pci_uninit+0x44/0x90 [hisi_qm]
| hisi_qm_uninit+0x28c/0x3b0 [hisi_qm]
| hpre_remove+0x16c/0x1c0 [hisi_hpre]
| pci_device_remove+0x6c/0x264
[...]
The reason for this call trace is that when the MSE is disabled, the value
of capability registers in the BAR space become invalid. This will make the
subsequent unregister process get the wrong irq vector through capability
registers and get the wrong irq number by pci_irq_vector().
So add a capability table structure to pre-store the valid value of the irq
information capability register in qm init process, avoid obtaining invalid
capability register value after the MSE is disabled.
Fixes: 3536cc55cada ("crypto: hisilicon/qm - support get device irq information from hardware registers")
Signed-off-by: Zhiqi Song <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
Extract a public function to set qm algs and remove
the similar code for setting qm algs in each module.
Signed-off-by: Wenkai Lin <[email protected]>
Signed-off-by: Hao Fang <[email protected]>
Signed-off-by: Zhiqi Song <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
1. Remove redundant array element, prevent the size obtained by
ARRAY_SIZE() from qm_log_hw_error is greater than actual size.
2. Add comments in function qm_set_vf_mse() and qm_cq_ctx_cfg()
to make it easier to understand.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The 'QM_INIT' and 'QM_CLOSE' status of qm and 'QP_INIT'
and 'QP_CLOSE' status of queue are not actually used. Currently,
driver only needs to switch status when the device or queue
is enabled or stopped, Therefore, remove unneeded status to
simplify driver. In addition, rename'QM_START to'QM_WORK' for
ease to understand.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
1. Remove unnecessary brackets in function hisi_acc_create_sgl_pool().
2. Modify local variable type, ensure that the variable type is
consistent with the variable type to be compared.
3. Because the function clear_hw_sgl_sge() is in the task process,
obtain the value of le16_to_cpu(hw_sgl->entry_sum_in_sgl) before
loop execting to shorten the loop execution time.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
To meet specific application scenarios, the function of switching between
the high performance mode and the high compression mode is added.
Use the perf_mode=0/1 configuration to set the compression high perf mode,
0(default, high compression mode), 1(high performance mode). These two
modes only apply to the compression direction and are compatible with
software algorithm in both directions.
Signed-off-by: Chenghai Huang <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
The 'offset' type is unsigned long in 'struct debugfs_reg32',
so type of values casts to unsigned long long is incorrect, and the
values do not require type cast, remove them.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
When device is abnormal and reports abnormal interrupt event to driver,
the driver can print device information for error analysis. This patch
adds some device error-related information output after the device reports
an abnormal interrupt.
Signed-off-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|
|
In the scenario where the accelerator business is fully loaded.
When the workqueue receiving messages and performing callback
processing, there are a large number of messages that need to be
received, and there are continuously messages that have been
processed and need to be received.
This will cause the receive loop here to be locked for a long time.
This scenario will cause watchdog timeout problems on OS with kernel
preemption turned off.
The error logs:
watchdog: BUG: soft lockup - CPU#23 stuck for 23s! [kworker/u262:1:1407]
[ 1461.978428][ C23] Call trace:
[ 1461.981890][ C23] complete+0x8c/0xf0
[ 1461.986031][ C23] kcryptd_async_done+0x154/0x1f4 [dm_crypt]
[ 1461.992154][ C23] sec_skcipher_callback+0x7c/0xf4 [hisi_sec2]
[ 1461.998446][ C23] sec_req_cb+0x104/0x1f4 [hisi_sec2]
[ 1462.003950][ C23] qm_poll_req_cb+0xcc/0x150 [hisi_qm]
[ 1462.009531][ C23] qm_work_process+0x60/0xc0 [hisi_qm]
[ 1462.015101][ C23] process_one_work+0x1c4/0x470
[ 1462.020052][ C23] worker_thread+0x150/0x3c4
[ 1462.024735][ C23] kthread+0x108/0x13c
[ 1462.028889][ C23] ret_from_fork+0x10/0x18
Therefore, it is necessary to add an actively scheduled operation in the
while loop to prevent this problem.
After adding it, no matter whether the OS turns on or off the kernel
preemption function. Neither will cause watchdog timeout issues.
Signed-off-by: Longfang Liu <[email protected]>
Signed-off-by: Herbert Xu <[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 ignored (apart
from emitting a warning) 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. Eventually after all drivers
are converted, .remove_new() will be renamed to .remove().
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]>
Reviewed-by: Weili Qian <[email protected]>
Signed-off-by: Herbert Xu <[email protected]>
|