- 06 Nov, 2019 18 commits
-
-
James Smart authored
The current driver attempts to allocate an interrupt vector per cpu using the systems managed IRQ allocator (flag PCI_IRQ_AFFINITY). The system IRQ allocator will either provide the per-cpu vector, or return fewer vectors. When fewer vectors, they are evenly spread between the numa nodes on the system. When run on an AMD architecture, if interrupts occur to a cpu that is not in the same numa node as the adapter generating the interrupt, there are extreme costs and overheads in performance. Thus, if 1:1 vector allocation is used, or the "balanced" vectors in the other numa nodes, performance can be hit significantly. A much more performant model is to allocate interrupts only on the cpus that are in the numa node where the adapter resides. I/O completion is still performed by the cpu where the I/O was generated. Unfortunately, there is no flag to request the managed IRQ subsystem allocate vectors only for the CPUs in the numa node as the adapter. On AMD architecture, revert the irq allocation to the normal style (non-managed) and then use irq_set_affinity_hint() to set the cpu affinity and disable user-space rebalancing. Tie the support into CPU offline/online. If the cpu being offlined owns a vector, the vector is re-affinitized to one of the other CPUs on the same numa node. If there are no more CPUs on the numa node, the vector has all affinity removed and lets the system determine where it's serviced. Similarly, when the cpu that owned a vector comes online, the vector is reaffinitized to the cpu. Link: https://lore.kernel.org/r/20191105005708.7399-10-jsmart2021@gmail.comSigned-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>
-
James Smart authored
The recent affinitization didn't address cpu offlining/onlining. If an interrupt vector is shared and the low order cpu owning the vector is offlined, as interrupts are managed, the vector is taken offline. This causes the other CPUs sharing the vector will hang as they can't get io completions. Correct by registering callbacks with the system for Offline/Online events. When a cpu is taken offline, its eq, which is tied to an interrupt vector is found. If the cpu is the "owner" of the vector and if the eq/vector is shared by other CPUs, the eq is placed into a polled mode. Additionally, code paths that perform io submission on the "sharing CPUs" will check the eq state and poll for completion after submission of new io to a wq that uses the eq. Similarly, when a cpu comes back online and owns an offlined vector, the eq is taken out of polled mode and rearmed to start driving interrupts for eq. Link: https://lore.kernel.org/r/20191105005708.7399-9-jsmart2021@gmail.comSigned-off-by:
-
James Smart authored
Current message on FAWWN events is rather cryptic. Expand the message to clarify its meaning. Link: https://lore.kernel.org/r/20191105005708.7399-8-jsmart2021@gmail.comSigned-off-by:
-
James Smart authored
Prior to the last FC-NVME-2 draft, SLER and CONF were independent. SLER now requires CONF to be set. Revise the NVME PRLI checking to look for both inorder to enable SLER. Link: https://lore.kernel.org/r/20191105005708.7399-7-jsmart2021@gmail.comSigned-off-by:
-
James Smart authored
The recently posted patch had a typo that incorrectly tested the receiving function. Fix the typo (change == to !=) Fixes: 95bfc6d8 ("scsi: lpfc: Make FW logging dynamically configurable") Link: https://lore.kernel.org/r/20191105005708.7399-6-jsmart2021@gmail.comSigned-off-by:
-
James Smart authored
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.comSigned-off-by:
-
James Smart authored
When reading sysfs nvme_info file while a remote port leaves and comes back, a NULL pointer is encountered. The issue is due to ndlp list corruption as the the nvme_info_show does not use the same lock as the rest of the code. Correct by removing the rcu_xxx_lock calls and replace by the host_lock and phba->hbaLock spinlocks that are used by the rest of the driver. Given we're called from sysfs, we are safe to use _irq rather than _irqsave. Link: https://lore.kernel.org/r/20191105005708.7399-4-jsmart2021@gmail.comSigned-off-by:
-
James Smart authored
The driver today is reading service parameters from the firmware and then overwriting the firmware-provided values with values of its own. There are some switch features that require preliminary FLOGI's that are switch-specific and done prior to the actual fabric FLOGI for traffic. The fw will perform those FLOGIs and will revise the service parameters for the features configured. As the driver later overwrites those values with its own values, it misconfigures things like BBSCN use by doing so. Correct by eliminating the driver-overwrite of firmware values. The driver correctly re-reads the service parameters after each link up to obtain the latest values from firmware. Link: https://lore.kernel.org/r/20191105005708.7399-3-jsmart2021@gmail.comSigned-off-by:
-
James Smart authored
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.comSigned-off-by:
-
Al Viro authored
simply not needed there - neither sg_new_read() nor sg_new_write() need it. Link: https://lore.kernel.org/r/20191017193925.25539-8-viro@ZenIV.linux.org.ukSigned-off-by:
Al Viro <viro@zeniv.linux.org.uk> Acked-by:
Douglas Gilbert <dgilbert@interlog.com> Signed-off-by:
-
Al Viro authored
Just use plain copy_from_user() and get_user(). Note that while a buf-derived pointer gets stored into ->dxferp, all places that actually use the resulting value feed it either to import_iovec() or to import_single_range(), and both will do validation. Link: https://lore.kernel.org/r/20191017193925.25539-7-viro@ZenIV.linux.org.ukSigned-off-by:
-
Al Viro authored
Use copy_..._user() instead, both in sg_read() and in sg_read_oxfer(). And don't open-code memdup_user()... Link: https://lore.kernel.org/r/20191017193925.25539-6-viro@ZenIV.linux.org.ukSigned-off-by:
-
Al Viro authored
... just use copy_from_user(). We copy only SZ_SG_IO_HDR bytes, so that would, strictly speaking, loosen the check. However, for call chains via ->write() the caller has actually checked the entire range and SG_IO passes exactly SZ_SG_IO_HDR for count. So no visible behaviour changes happen if we check only what we really need for copyin. Link: https://lore.kernel.org/r/20191017193925.25539-5-viro@ZenIV.linux.org.ukSigned-off-by:
-
Al Viro authored
We don't need to allocate a temporary buffer and read the entire structure in it, only to fetch a single field and free what we'd allocated. Just use get_user() and be done with it... Link: https://lore.kernel.org/r/20191017193925.25539-4-viro@ZenIV.linux.org.ukSigned-off-by:
-
Al Viro authored
Link: https://lore.kernel.org/r/20191017193925.25539-3-viro@ZenIV.linux.org.ukSigned-off-by:
-
Al Viro authored
Link: https://lore.kernel.org/r/20191017193925.25539-2-viro@ZenIV.linux.org.ukSigned-off-by:
-
Al Viro authored
First of all, __put_user() can fail with access_ok() succeeding. And access_ok() + __copy_to_user() is spelled copy_to_user()... __put_user() *can* fail with access_ok() succeeding... Link: https://lore.kernel.org/r/20191017193925.25539-1-viro@ZenIV.linux.org.ukSigned-off-by:
-
Pan Bian authored
The memory chunk io_req is released by mempool_free. Accessing io_req->start_time will result in a use after free bug. The variable start_time is a backup of the timestamp. So, use start_time here to avoid use after free. Link: https://lore.kernel.org/r/1572881182-37664-1-git-send-email-bianpan2016@163.comSigned-off-by:
Pan Bian <bianpan2016@163.com> Reviewed-by:
Satish Kharat <satishkh@cisco.com> Signed-off-by:
-
- 02 Nov, 2019 5 commits
-
-
Bart Van Assche authored
This patch fixes an unintended sign extension on left shifts. From Colin King: "Shifting a u8 left will cause the value to be promoted to an integer. If the top bit of the u8 is set then the following conversion to an u64 will sign extend the value causing the upper 32 bits to be set in the result." Fix this by using get_unaligned_be*() instead. Fixes: bf816235 ("[SCSI] add scsi trace core functions and put trace points") Cc: Christoph Hellwig <hch@lst.de> Cc: Hannes Reinecke <hare@suse.com> Cc: Douglas Gilbert <dgilbert@interlog.com> Link: https://lore.kernel.org/r/20191101211447.187151-1-bvanassche@acm.orgReported-by:
Colin Ian King <colin.king@canonical.com> Signed-off-by:
Bart Van Assche <bvanassche@acm.org> Signed-off-by:
-
Saurav Girepunje authored
Only csio_hw_free() calling csio_dfs_destroy() and it is not checking return value. So remove the return from csio_dfs_destroy(). Link: https://lore.kernel.org/r/20191028194234.GA27848@sauravSigned-off-by:
Saurav Girepunje <saurav.girepunje@gmail.com> Signed-off-by:
-
Saurav Girepunje authored
debugfs_remove_recursive() has taken the null pointer into account. Remove the null check before debugfs_remove_recursive(). Link: https://lore.kernel.org/r/20191026195625.GA22455@sauravSigned-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by:
-
Saurav Girepunje authored
Replace assignment of 0 to pointer with NULL assignment. Link: https://lore.kernel.org/r/20191025135010.GA6191@sauravSigned-off-by:
Jack Wang <jinpu.wang@cloud.ionos.com> Signed-off-by:
-
Ming Lei authored
It isn't necessary to check the host depth in scsi_queue_rq() any more since it has been respected by blk-mq before calling scsi_queue_rq() via getting driver tag. Lots of LUNs may attach to same host and per-host IOPS may reach millions, so we should avoid expensive atomic operations on the host-wide counter in the IO path. This patch implements scsi_host_busy() via blk_mq_tagset_busy_iter() with one scsi command state for reading the count of busy IOs for scsi_mq. It is observed that IOPS is increased by 15% in IO test on scsi_debug (32 LUNs, 32 submit queues, 1024 can_queue, libaio/dio) in a dual-socket system. Cc: Jens Axboe <axboe@kernel.dk> Cc: Ewan D. Milne <emilne@redhat.com> Cc: Omar Sandoval <osandov@fb.com>, Cc: "Martin K. Petersen" <martin.petersen@oracle.com>, Cc: James Bottomley <james.bottomley@hansenpartnership.com>, Cc: Christoph Hellwig <hch@lst.de>, Cc: Kashyap Desai <kashyap.desai@broadcom.com> Cc: Hannes Reinecke <hare@suse.de> Cc: Laurence Oberman <loberman@redhat.com> Cc: Bart Van Assche <bvanassche@acm.org> Link: https://lore.kernel.org/r/20191025065855.6309-1-ming.lei@redhat.comSigned-off-by:
Ming Lei <ming.lei@redhat.com> Reviewed-by:
Jens Axboe <axboe@kernel.dk> Reviewed-by:
-
- 01 Nov, 2019 3 commits
-
-
Bart Van Assche authored
Declare all variables that hold dev_cmd_type values as an enum instead of as an int. Cc: Yaniv Gardi <ygardi@codeaurora.org> Cc: Subhash Jadavani <subhashj@codeaurora.org> Cc: Stanley Chu <stanley.chu@mediatek.com> Cc: Avri Altman <avri.altman@wdc.com> Cc: Tomas Winkler <tomas.winkler@intel.com> Link: https://lore.kernel.org/r/20191029230710.211926-3-bvanassche@acm.orgSigned-off-by:
Avri Altman <avri.altman@wdc.com> Signed-off-by:
-
Bart Van Assche authored
Fix the following three kernel-doc warnings: drivers/scsi/ufs/ufs_bsg.c:165: warning: Function parameter or member 'hba' not described in 'ufs_bsg_remove' drivers/scsi/ufs/ufshcd.c:5789: warning: Function parameter or member 'cmd_type' not described in 'ufshcd_issue_devman_upiu_cmd' drivers/scsi/ufs/ufshcd.c:5789: warning: Excess function parameter 'msgcode' description in 'ufshcd_issue_devman_upiu_cmd' Cc: Yaniv Gardi <ygardi@codeaurora.org> Cc: Subhash Jadavani <subhashj@codeaurora.org> Cc: Stanley Chu <stanley.chu@mediatek.com> Cc: Avri Altman <avri.altman@wdc.com> Cc: Tomas Winkler <tomas.winkler@intel.com> Link: https://lore.kernel.org/r/20191029230710.211926-2-bvanassche@acm.orgSigned-off-by:
-
Bean Huo authored
There is no need to call ufshcd_def_desc_sizes() in ufshcd_init(), since descriptor lengths will be checked and initialized later in ufshcd_init_desc_sizes(). Fixes: a4b0e8a4(scsi: ufs: Factor out ufshcd_read_desc_param) Link: https://lore.kernel.org/r/BN7PR08MB5684A3ACE214C3D4792CE729DB610@BN7PR08MB5684.namprd08.prod.outlook.comSigned-off-by:
Bean Huo <beanhuo@micron.com> Acked-by: Avri Altman <avri.altman.wdc.com> Reviewed-by:
Can Guo <cang@codeaurora.org> Signed-off-by:
-
- 29 Oct, 2019 14 commits
-
-
Steffen Maier authored
While v2.6.26 commit b75db731 ("[SCSI] zfcp: Add qtcb dump to hba debug trace") is right that we don't want to flood the (payload) trace ring buffer, we don't trace successful FCP command responses by default. So we can include the channel log for problem determination with failed responses of any FSF request type. Fixes: b75db731 ("[SCSI] zfcp: Add qtcb dump to hba debug trace") Fixes: a54ca0f6 ("[SCSI] zfcp: Redesign of the debug tracing for HBA records.") Cc: <stable@vger.kernel.org> #2.6.38+ Link: https://lore.kernel.org/r/e37597b5c4ae123aaa85fd86c23a9f71e994e4a9.1572018132.git.bblock@linux.ibm.comReviewed-by:
Benjamin Block <bblock@linux.ibm.com> Signed-off-by:
Steffen Maier <maier@linux.ibm.com> Signed-off-by:
-
Steffen Maier authored
No functional change. The unary not operator only applies to the sub expression before the logical or. So we return early if (not running) or failed. Link: https://lore.kernel.org/r/df4f897f6e83eaa528465d0858d5a22daac47a2f.1572018132.git.bblock@linux.ibm.comReviewed-by:
Jens Remus <jremus@linux.ibm.com> Reviewed-by:
-
Benjamin Block authored
Replace the static define (ZFCP_DIAG_MAX_AGE) with a per-adapter variable (${adapter}->diagnostics->max_age). This new variable is exported via sysfs, along with other, already existing adapter variables, and can both be read and written. This way users can choose how much time should pass between refreshes of diagnostic buffers. The default value for the age remains to be five seconds. By setting this new variable to 0, the caching of diagnostic buffers for userspace accesses can also be completely removed. All diagnostic buffers of a given adapter are subject to this setting in the same way. Link: https://lore.kernel.org/r/b1d0977cc884b16dd4ca6418e4320c56a4c31d63.1572018132.git.bblock@linux.ibm.comReviewed-by: -
Benjamin Block authored
Adds implicit updates of cached diagnostics via Exchange Config Data when reading sysfs attributes interfacing them. Right now this only affects the new B2B-Credit diagnostic attribute. This uses the same mechanism previously also used for cached diagnostics of Exchange Port Data. Link: https://lore.kernel.org/r/60a94f55f2630b74b468fed5f39880208abb2679.1572018132.git.bblock@linux.ibm.comReviewed-by:
-
Benjamin Block authored
In addition to the diagnostic data from the local SFP transceiver this patch adds an interface to read the advertised buffer-to-buffer credit from the local FC_Port. With this patch the userspace-interface will only read data stored in the corresponding "diagnostic buffer" (that was stored during completion of a previous Exchange Config Data command). Implicit updating will follow later in this series. Link: https://lore.kernel.org/r/8a53aef87b53c50cfb1a3425b799bacb6f82b832.1572018132.git.bblock@linux.ibm.comReviewed-by:
-
Benjamin Block authored
This patch adds implicit updates to the sysfs entries that read the diagnostic data stored in the "caching buffer" for Exchange Port Data. An update is triggered once the buffer is older than ZFCP_DIAG_MAX_AGE milliseconds (5s). This entails sending an Exchange Port Data command to the FCP-Channel, and during its ingress path updating the cached data and the timestamp. To prevent multiple concurrent userspace-applications from triggering this update in parallel we synchronize all of them using a wait-queue (waiting threads are interruptible; the updating thread is not). Link: https://lore.kernel.org/r/c145b5cfc99a63b6a018b1184fbd27bb09c955f5.1572018132.git.bblock@linux.ibm.comReviewed-by:
-
Benjamin Block authored
This adds an interface to read the diagnostics of the local SFP transceiver of an FCP-Channel from userspace. This comes in the form of new sysfs entries that are attached to the CCW device representing the FCP device. Each type of data gets its own sysfs entry; the whole collection of entries is pooled into a new child-directory of the CCW device node: "diagnostics". Adds sysfs entries for: * sfp_invalid: boolean value evaluating to whether the following 5 fields are invalid; {0, 1}; 1 - invalid * temperature: transceiver temp.; unit 1/256°C; range [-128°C, +128°C] * vcc: supply voltage; unit 100μV; range [0, 6.55V] * tx_bias: transmitter laser bias current; unit 2μA; range [0, 131mA] * tx_power: coupled TX output power; unit 0.1μW; range [0, 6.5mW] * rx_power: received optical power; unit 0.1μW; range [0, 6.5mW] * optical_port: boolean value evaluating to whether the FCP-Channel has an optical port; {0, 1}; 1 - optical * fec_active: boolean value evaluating to whether 16G FEC is active; {0, 1}; 1 - active * port_tx_type: nibble describing the port type; {0, 1, 2, 3}; 0 - unknown, 1 - short wave, 2 - long wave LC 1310nm, 3 - long wave LL 1550nm * connector_type: two bits describing the connector type; {0, 1}; 0 - unknown, 1 - SFP+ This is only supported if the FCP-Channel in turn supports reporting the SFP Diagnostic Data, otherwise read() on these new entries will return EOPNOTSUPP (this affects only adapters older than FICON Express8S, on Mainframe generations older than z14). Other possible errors for read() include ENOLINK, ENODEV and ENOMEM. With this patch the userspace-interface will only read data stored in the corresponding "diagnostic buffer" (that was stored during completion of an previous Exchange Port Data command). Implicit updating will follow later in this series. Link: https://lore.kernel.org/r/1f9cce7c829c881e7d71a3f10c5b57f3dd84ab32.1572018132.git.bblock@linux.ibm.comReviewed-by: -
Benjamin Block authored
A new FCP channel feature allows us to read the diagnostics from our local SFP transceivers. To make use of that add a flag (FSF_FEATURE_REQUEST_SFP_DATA) to the feature-set we request from the FCP channel. Whether the channel actually implements this can be determined via an other new flag (FSF_FEATURE_REPORT_SFP_DATA), that is set in the adapter_features field of the adapter structure after Exchange Config Data finished. Also add the corresponding definitions in the QTCB Bottom for Exchange Port Data. These new definitions are only valid, if FSF_FEATURE_REPORT_SFP_DATA is set. Link: https://lore.kernel.org/r/ee1eba4de71eb06b4d82207ad4f428429346156f.1572018132.git.bblock@linux.ibm.comReviewed-by:
-
Benjamin Block authored
In the same vein as the previous patch, add diagnostic data capture for the Exchange Config Data command. Link: https://lore.kernel.org/r/7d8ac0a6cad403fa8f8b888693476a84e80a277b.1572018131.git.bblock@linux.ibm.comReviewed-by:
-
Benjamin Block authored
The FCP channel exposes two central interfaces to receive information about the local FCP-Adapter/-Port: Exchange Port and Exchange Config Data. Using these commands can negatively impact the adapter if we allow them to be sent at a very high rate. The later parts of this patchset will introduce new user-interfaces to receive more diagnostics from the adapter. To prevent any negative impact from using those, this patch adds a simple caching-mechanism that will prevent a malicious/faulty userspace-application from generating an abnormal high amount of Exchange Port/Config Data traffic. Relevant diagnostic data that is received via Exchange Config/Port Data is cached in buffers associated with the corresponding adapter-struct. Each buffer is associated with a timestamp that signals how old the data is, and, added via a following patch in this series, lets userspace-interfaces determine when the data is too old and needs to be updated. Buffer-updates are made during the normal response path of the corresponding command. With this patch only the output of the Exchange Port Data command is captured. Link: https://lore.kernel.org/r/054ca020ce0a53dc0d9176428bea373898944e6a.1572018130.git.bblock@linux.ibm.comReviewed-by:
-
Benjamin Block authored
Adds a new FSF-Request status flag (ZFCP_STATUS_FSFREQ_XDATAINCOMPLETE) that signal that the data received using Exchange Config Data or Exchange Port Data was incomplete. This new flags is set in the respective handlers during the response path. With this patch, only the synchronous FSF-functions for each command got support for the new flag, otherwise it is transparent. Together with this new flag and already existing status flags the synchronous FSF-functions are extended to now detect whether the received data is complete, incomplete or completely invalid (this includes cases where a command ran into a timeout). This is now signaled back to the caller, where previously only failures on the request path would result in a bad return-code. For complete data the return-code remains 0. For incomplete data a new return-code -EAGAIN is added to the function-interface. For completely invalid data the already existing return-code -EIO is reused - formerly this was used to signal failures on the request path. Existing callers of the FSF-functions are adjusted so that they behave as before for return-code 0 and -EAGAIN, to not change the user-interface. As -EIO existed all along, it was already exposed to the user - and needed handling - and will now also be exposed in this new special case. Link: https://lore.kernel.org/r/e14f0702fa2b00a4d1f37c7981a13f2dd1ea2c83.1572018130.git.bblock@linux.ibm.comReviewed-by:
-
YueHaibing authored
Fix sparse warning: drivers/scsi/lpfc/lpfc_debugfs.c:2083:1: warning: symbol 'lpfc_debugfs_ras_log_data' was not declared. Should it be static? Link: https://lore.kernel.org/r/20191028132556.16272-1-yuehaibing@huawei.comReported-by:
Hulk Robot <hulkci@huawei.com> Signed-off-by:
YueHaibing <yuehaibing@huawei.com> Reviewed-by:
James Smart <james.smart@broadcom.com> Signed-off-by:
-
Saurav Girepunje authored
mempool_destroy has taken null pointer check into account. Remove the redundant check. Link: https://lore.kernel.org/r/20191026194712.GA22249@sauravSigned-off-by:
-
James Smart authored
convert MAGIC_NUMER_xxx to MAGIC_NUMBER_xxx Link: https://lore.kernel.org/r/20191025184342.6623-1-jsmart2021@gmail.comSigned-off-by:
Ewan D. Milne <emilne@redhat.com> Signed-off-by:
-
