- 07 Dec, 2016 1 commit
-
-
Oleg Drokin authored
I have been having a lot of unexplainable crashes in osc_lru_shrink lately that I could not see a good explanation for and then I found this patch that slip under the radar somehow that incorrectly converted while loop for lru list iteration into list_for_each_entry_safe totally ignoring that in the body of the loop we drop spinlocks guarding this list and move list entries around. Not sure why it was not showing up right away, perhaps some of the more recent LRU changes committed caused some extra pressure on this code that finally highlighted the breakage. Reverts: 8adddc36 ("staging: lustre: osc: Use list_for_each_entry_safe") CC: Bhaktipriya Shridhar <bhaktipriya96@gmail.com> Signed-off-by:
Oleg Drokin <green@linuxhacker.ru> Cc: stable <stable@vger.kernel.org> # 4.6+ Signed-off-by:
Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-
- 06 Dec, 2016 39 commits
-
-
Greg Kroah-Hartman authored
A "real" driver for this hardware has now landed in the networking tree, so remove this old staging driver so that we don't have multiple drivers for the same hardware, and so people don't waste their time trying to clean up this old code. Cc: Lior Dotan <liodot@gmail.com> Cc: Christopher Harrer <charrer@alacritech.com> Cc: David Miller <davem@davemloft.net> Signed-off-by: -
Alexander Zarochentsev authored
Removing lnet upcall infrastructure completely as nobody uses it anymore. The upcall causes a delay before calling BUG() and might even cause a hang making getting a crash dump unreliable or containing outdated info. Signed-off-by:
Alexander Zarochentsev <alexander.zarochentsev@seagate.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-8418 Seagate-bug-id: MRP-2939 Reviewed-on: http://review.whamcloud.com/21440Reviewed-by:
Andreas Dilger <andreas.dilger@intel.com> Reviewed-by:
James Simmons <uja.ornl@yahoo.com> Reviewed-by:
Oleg Drokin <oleg.drokin@intel.com> Signed-off-by:
James Simmons <jsimmons@infradead.org> Signed-off-by:
-
Yang Sheng authored
Remove set but unused variables in nidstring.c and osc_request.c as reported by make W=1. Signed-off-by:
Yang Sheng <yang.sheng@intel.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-8378 Reviewed-on: http://review.whamcloud.com/23221Reviewed-by:
Bob Glossman <bob.glossman@intel.com> Reviewed-by:
Emoly Liu <emoly.liu@intel.com> Reviewed-by:
-
Jinshan Xiong authored
If osc_io_readahead() finds a lock that belongs to the previous instance of osc_object, the lock data pointer will be null. It has to instantiate with new instance otherwise those pages won't be destroyed at lock cancel, and then finally hit the assertion in osc_req_attr_set(). This patch revised dlmlock_at_pgoff() to call osc_match_base() to find caching locks for readahead. And new osc_object will be set to the lock if it doesn't have one yet. Signed-off-by:
Jinshan Xiong <jinshan.xiong@intel.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-8005 Reviewed-on: http://review.whamcloud.com/19453Reviewed-by:
Bobi Jam <bobijam@hotmail.com> Reviewed-by:
John L. Hammond <john.hammond@intel.com> Reviewed-by:
-
Mikhal Pershin authored
The import connect flags might be cleared by ptlrpc_connect_import() wrongly if there is still connect interpret function is running. Use imp_connected boolean variable to indicate that we are still interpretting connect reply and don't try to reconnect until it ends. Signed-off-by:
Mikhal Pershin <mike.pershin@intel.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7558 Reviewed-on: http://review.whamcloud.com/19312Reviewed-by:
-
Bobi Jam authored
In fault IO initialization, inode's mtime is saved, and after getting locks, when the IO is about to start, vvp_io_fault_start() checks the mtime's intactness. It's a false alarm, since the timestamp from MDS could be stale, we maintain mtime mainly on OST objects, and if the check in vvp_io_fault_start() happens before mtime on OST objects are merged, it will get wrong timestamp from the inode, even the timestamp it fetched in vvp_io_fault_init() could be wrong in the first place. This patch remove the mtime check in vvp_io_fault_start(). Signed-off-by:
Bobi Jam <bobijam.xu@intel.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7198 Reviewed-on: http://review.whamcloud.com/19162Reviewed-by:
-
Yang Sheng authored
Only update file times if page_mkwrite is not set. So we need call file_update_time by ourselves. Signed-off-by:
Niu Yawei <yawei.niu@intel.com> Reviewed-by:
-
Jinshan Xiong authored
To make the ptlrpc be able to size 16MB IO Signed-off-by:
Gu Zheng <gzheng@ddn.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7990 Reviewed-on: http://review.whamcloud.com/19366Reviewed-by:
-
Fan Yong authored
It is the sponsor thread of the statahead thread to update the sai::sai_index_wait. Originally, it didn't hold the lli_sa_lock when did that. Becuase of out-of-order execution others may miss to wakeup such thread. On the other hand, if the statahead RPC gets failure, it should wakeup the sponsor thread, not the statahead thread. Signed-off-by:
Li Xi <lixi@ddn.com> Signed-off-by:
Fan Yong <fan.yong@intel.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7828 Reviewed-on: http://review.whamcloud.com/18499Reviewed-by:
Lai Siyao <lai.siyao@intel.com> Reviewed-by:
-
Li Dongyang authored
Allocating a big hash table using the current formula does not really work for clients. We will create new hash table for each mount on a single client which is a lot of memory more than expected. This patch limits the hash table up to 8M for clients, which has 524288 entries. Signed-off-by:
Li Dongyang <dongyang.li@anu.edu.au> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7689 Reviewed-on: http://review.whamcloud.com/18048Reviewed-by:
Alex Zhuravlev <alexey.zhuravlev@intel.com> Reviewed-by:
-
Wally Wang authored
When Lustre servers enable 'suppress_pings', all clients will stop pinging. However, some clients may not have external mechanism to notify Lustre servers for node death and therefore need to preserve the Lustre ping. This patch provides a mount option 'always_ping' so that the client will not stop pinging even if the server has enabled 'suppress_pings'. Signed-off-by:
Wally Wang <wang@cray.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-6391 Reviewed-on: http://review.whamcloud.com/14127Reviewed-by:
Li Wei <wei.g.li@intel.com> Reviewed-by:
Chris Horn <hornc@cray.com> Reviewed-by:
-
Jinshan Xiong authored
ras_window_len should only be updated in ras_update() by read pattern and it can't be adjusted in ll_readahead() at all; ras_consecutive_pages is used to detect read pattern from mmap. It will be used to increase read ahead window length gradually. Signed-off-by:
-
Alex Zhuravlev authored
if the callers wants to purge all objects, then scanning should start from the first bucket. Signed-off-by:
Faccini Bruno <bruno.faccini@intel.com> Reviewed-by:
-
Parinay Kondekar authored
In ll_dir_ioctl() two identical comparisions are present for return code (rc) of ll_dir_getstripe(). This patch removes the other inside if( ) condition which is not necessary. Signed-off-by:
Parinay Kondekar <parinay.kondekar@seagate.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-6512 Reviewed-on: http://review.whamcloud.com/18027Reviewed-by:
-
Yang Sheng authored
Add more debugging info. Signed-off-by:
-
Li Dongyang authored
O_WRONLY/O_RDWR open on a file will get EROFS on a read only client, but the rpc gets sent to the mdt anyway. mdt will increase the mot_write_count of the mdt object, blocking subsequent FMODE_EXEC open to the same file. This patch makes sure we fail the FMODE_WRITE open with EROFS on the client straight away without sending the rpc to mdt. Signed-off-by:
Ian Costello <icostello@ddn.com> Reviewed-by:
Nathaniel Clark <nathaniel.l.clark@intel.com> Reviewed-by:
-
wang di authored
Once update request fails due to eviction or other failures, all of update request in the sending list should return fail, because after the failure, the update log in the following request will have wrong llog bitmap. Signed-off-by:
wang di <di.wang@intel.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7039 Reviewed-on: http://review.whamcloud.com/16969Reviewed-by:
-
wang di authored
During intent open, it was found that if the parent has been migrated to another MDT, it should retry the open request with the new object, so it needs to keep the old object in the orphan list, which will be cleanup during next recovery. Note: if the client still using the old FID after next recovery, it will return -ENOENT for the application. Also enqueue the lease lock of the migrating file, then compare the lease before migration to make sure no other clients open the file at the same time. Signed-off-by:
-
Ashish Purkar authored
Corrected newline specifier in debug log message. Signed-off-by:
Ashish Purkar <ashish.purkar@seagate.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-7029 Reviewed-on: http://review.whamcloud.com/16046Reviewed-by:
-
Jinshan Xiong authored
This flag should be cleared atomically after the op_data flag MDS_DATA_MODIFIED is packed. Otherwise, if there exists an operation to dirty the file again, the state may be missed on the MDT. Stop using spin lock lli_lock to protect operations of changing file flags; using bit operations instead. Signed-off-by:
-
Dan Carpenter authored
We want to "sever" all the ways to get a new pointer to "pg". Signed-off-by:
Dan Carpenter <dan.carpenter@oracle.com> Signed-off-by:
-
Sandeep Jain authored
Fixes following Sparse errors. lprocfs_status.c:1568:5: error: symbol 'lprocfs_wr_root_squash' redeclared with different type... lprocfs_status.c:1632:5: error: symbol 'lprocfs_wr_nosquash_nids' redeclared with different type... Signed-off-by:
Sandeep Jain <sandeepjain.linux@gmail.com> Acked-by:
-
Sandeep Jain authored
Fix following sparse warning. mgc_request.c:376:1: warning: symbol 'llog_process_lock' was not declared. Should it be static? Signed-off-by:
-
Pan Bian authored
At the end of function lstcon_group_info(), "return 0" seems improper. It may be better to return the value of rc. Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=188811Signed-off-by:
Pan Bian <bianpan2016@163.com> Acked-by:
-
Jinshan Xiong authored
Early launch page LRU work in osc_io_rw_iter_init(); Change the page LRU shrinking policy by OSC attributes; Delete the contented lock osc_object::oo_seatbelt Signed-off-by:
-
Jinshan Xiong authored
Create the inline function cli_name() to get the name of the OSC device. Signed-off-by:
-
Jinshan Xiong authored
Add new information about the fields in struct client_obd. Signed-off-by:
-
Ben Evans authored
It was found if you sort the headers alphabetically that it reduced patch conflicts. This patch sorts the headers alphabetically and also place linux header first, then uapi header and finally the lustre kernel headers. Signed-off-by:
Ben Evans <bevans@cray.com> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-6401 Reviewed-on: http://review.whamcloud.com/16339Reviewed-by:
-
Ben Evans authored
Create headers for pack_generic.c and llog_swab.c Reference only where needed. This separates out the kernel only code from lustre_idl.h that is an UAPI header. Signed-off-by:
-
Manoj Sawai authored
Fixed 2 coding style errors about trailing whitespaces. Signed-off-by:
Manoj Sawai <mas@iitkgp.ac.in> Signed-off-by:
-
Yamanappagouda Patil authored
Fixed checkpatch.pl errors related to "space prohibited after that '*' or '&'" in ks_wlan_net.c file. Signed-off-by:
Yamanappagouda Patil <goudapatilk@gmail.com> Signed-off-by:
-
Yamanappagouda Patil authored
Fixed checkpatch.pl warnings 'Missing blank line after declaration' in ks_wlan_net.c file. Signed-off-by:
-
Guillaume Tucker authored
Clean up the code in inline functions that deal with page and range addresses. Use bool instead of int for boolean return types and remove superfluous brackets. Signed-off-by:
Guillaume Tucker <guillaume.tucker@gmail.com> Signed-off-by:
-
Guillaume Tucker authored
Convert range_size and range_on_lru macros to inline functions to fix checkpatch check: CHECK: Macro argument reuse 'range' - possible side-effects? Signed-off-by:
-
Yan Laijun authored
Fixed checkpatch warning "line over 80 characters" in wlan-ng/hfa384x_usb.c file. Signed-off-by:
Yan Laijun <yan.laijun@gmail.com> Signed-off-by:
-
PrasannaKumar Muralidharan authored
Add spaces around + for better readability of the code. Signed-off-by:
PrasannaKumar Muralidharan <prasannatsmkumar@gmail.com> Signed-off-by:
-
Harsh Jain authored
Fixes following checkpatch.pl warning Block comments should align the * on each line Signed-off-by:
Harsh Jain <harshjain32@gmail.com> Signed-off-by:
-
Fernando Apesteguia authored
This patch fixes a checkpatch warning. Signed-off-by:
Fernando Apesteguia <fernando.apesteguia@gmail.com> Signed-off-by:
-
Cheah Kok Cheong authored
Add init/exit function to follow LKM semantics. Apparently this module can still load/unload without the init/exit function. Tested loading/unloading with and without this patch. Signed-off-by:
Cheah Kok Cheong <thrust73@gmail.com> Reviewed-by:
Ian Abbott <abbotti@mev.co.uk> Signed-off-by:
-
