Commit Graph

1614 Commits

Author SHA1 Message Date
Thomas Schoebel-Theuer 48654bffea infra: add script rework-mars-for-upstream.pl 2015-12-31 10:38:57 +01:00
Thomas Schoebel-Theuer 0e2b7f1445 if: don't rely on total_len 2015-12-31 10:38:51 +01:00
Thomas Schoebel-Theuer 2633a24d67 if: adapt to blk_set_stacking_limits 2015-12-31 10:38:20 +01:00
Thomas Schoebel-Theuer 63b47b60bf if: adapt to presence of REQ_WRITE_SAME 2015-12-31 10:38:20 +01:00
Thomas Schoebel-Theuer 6769b65da4 if: also for blk_queue_max_segments 2015-12-31 10:38:20 +01:00
Thomas Schoebel-Theuer c70bd21638 if: new sector size conventions 2015-12-31 10:38:20 +01:00
Thomas Schoebel-Theuer a08ff3a255 if: adapt to changes, use blk_queue_split() 2015-12-31 10:29:30 +01:00
Thomas Schoebel-Theuer 6c3439e126 if: adapt to changes in bio_flagged 2015-12-31 10:27:39 +01:00
Thomas Schoebel-Theuer 80d541b079 aio: adapt to changes in get_unused_fd()
Only relevant for the out-of-tree version.

The AIO stuff needs to be re-implemented anyway.
2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer b7fe08ecb8 if: mark USE_CONGESTED_FN for upstream removal 2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer c978402578 if: mark USE_MAX_* for upstream removal 2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer a06a73f89d if: mark WB_STAT_BATCH dependency for upstream removal 2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer 96db69e2d3 if: explain reason for missing kmap() 2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer eef84829fe if: let DENY_READA go upstream
it hints that a better implementation should be done in future.
2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer 7977aecf74 if: mark statistics #ifdefs for upstream removal 2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer 349ac32d09 if: mark unplug_fn for upstream removal 2015-12-31 10:26:04 +01:00
Thomas Schoebel-Theuer 32038bb6b8 if: adapt to kernel 4.4 BLK_QC_T_NONE
see dece16353ef47d8d33f5302bc158072a9d65e26f
2015-12-31 10:26:03 +01:00
Thomas Schoebel-Theuer 8002a30373 if: remove kick_sem
This was very old provisionary code from the first prototype in 2010.
It was leading to the following deadlock on newer kernels:

Oct  1 02:44:02 istore-test-bs1 kernel: : INFO: task umount:1040 blocked for more than 120 seconds.
Oct  1 02:44:02 istore-test-bs1 kernel: : Tainted: G         C      4.3.0-rc3-next-20150929+ #6
Oct  1 02:44:02 istore-test-bs1 kernel: : "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
Oct  1 02:44:02 istore-test-bs1 kernel: : umount          D ffff880313164300     0  1040    919 0x00000000
Oct  1 02:44:02 istore-test-bs1 kernel: : ffff8801b3c979d8 0000000000000082 ffff880313164300 ffff8802fe8c6880
Oct  1 02:44:02 istore-test-bs1 kernel: : ffff8801b3c98000 ffff88030fec7650 7fffffffffffffff 0000000000000002
Oct  1 02:44:02 istore-test-bs1 kernel: : 0000000000000000 ffff8801b3c979f0 ffffffff8140917d 7fffffffffffffff
Oct  1 02:44:02 istore-test-bs1 kernel: : Call Trace:
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140917d>] schedule+0x76/0x8b
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140c1fc>] schedule_timeout+0x24/0xe0
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140c945>] ? _raw_spin_unlock_irq+0x27/0x32
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff81066dcd>] ? trace_hardirqs_on_caller+0x171/0x18d
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff81066df6>] ? trace_hardirqs_on+0xd/0xf
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140c945>] ? _raw_spin_unlock_irq+0x27/0x32
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140b824>] __down_common+0x8e/0xcf
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140b87d>] __down+0x18/0x1a
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff81064097>] down+0x28/0x38
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffffa03b1493>] if_make_request+0x2d4/0x93d [mars]
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8140c899>] ? _raw_spin_unlock+0x22/0x2c
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff8120c93b>] ? create_task_io_context+0x11b/0x12e
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff812083b4>] ? create_io_context+0x51/0x66
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff812086c1>] generic_make_request+0x97/0xd6
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff812087eb>] submit_bio+0xeb/0xf6
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff811ba82f>] _xfs_buf_ioapply+0x2c5/0x327
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff811bb65e>] xfs_buf_submit+0x68/0x9b
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff811d1aa9>] xlog_bdstrat+0x48/0x4f
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff811d35dc>] xlog_sync+0x269/0x339
Oct  1 02:44:02 istore-test-bs1 kernel: : [<ffffffff811d373c>] xlog_state_release_iclog+0x90/0xa0
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff811d4139>] _xfs_log_force+0xfb/0x1bc
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff811d4209>] xfs_log_force+0xf/0x30
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff811cd804>] xfs_fs_sync_fs+0x1d/0x3b
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810fbedc>] sync_filesystem+0x7a/0x8e
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810d96c7>] generic_shutdown_super+0x22/0xe5
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810d9921>] kill_block_super+0x22/0x65
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810d9a09>] deactivate_locked_super+0x36/0x63
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810d9d27>] deactivate_super+0x31/0x34
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810f0422>] cleanup_mnt+0x53/0x71
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810f0474>] __cleanup_mnt+0xd/0xf
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff81050974>] task_work_run+0x62/0x8e
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff8100105f>] prepare_exit_to_usermode+0x8c/0x9d
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff810010f2>] syscall_return_slowpath+0x82/0x8c
Oct  1 02:44:03 istore-test-bs1 kernel: : [<ffffffff8140d0b1>] int_ret_from_sys_call+0x25/0x9f
Oct  1 02:44:03 istore-test-bs1 kernel: : 1 lock held by umount/1040:
Oct  1 02:44:03 istore-test-bs1 kernel: : #0:  (&type->s_umount_key#24){++++++}, at: [<ffffffff810d9d1f>] deactivate_super+0x29/0x34
2015-12-31 10:11:45 +01:00
Thomas Schoebel-Theuer 731d0f0b06 infra: fix spelling 2015-12-31 10:08:32 +01:00
Thomas Schoebel-Theuer 454420d849 logger: fix LOGICAL_CONTINUATIONS 2015-12-31 10:08:32 +01:00
Thomas Schoebel-Theuer bc2f3d146f client: keep checkpatch AVOID_EXTERNS happy 2015-12-31 10:07:48 +01:00
Thomas Schoebel-Theuer 413cd5935a sio: correct whitespace 2015-12-31 10:07:48 +01:00
Thomas Schoebel-Theuer 87e782bace bio: correct spelling 2015-12-31 10:07:48 +01:00
Thomas Schoebel-Theuer b46e51d82e net: keep checkpatch AVOID_EXTERNS happy 2015-12-31 10:07:48 +01:00
Thomas Schoebel-Theuer c919eef44a bio: adapt bio_endio() to kernel 4.3 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 7776857384 if: adapt bio_endio() to kernel 4.3 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 46b9f4bab0 infra: adapt compat layer to kernel upstream 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 86e58c9138 if: remove misleading comment
The comment was conceptually wrong / outdated:
the XIO infrastructure _can_ already so READAhead simply by just ignoring
callbacks, in addition to the already present background IO priority.
2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer db846ec685 server: split into generic and strategy part
This is a preparation of instance orientation.
Fully dynamic separation not yet done.
Currently the separation is at linker level only.
But the different parts may go to different locations later.
2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer fb6af90bcd infra: make architectural dependency generic 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 12a8d09b11 infra: lib_rank: remove superfluous definition 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer fb0c52538e infra: lib_limiter: remove dependency from MARS and BRICK 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 1c7b01b271 infra: lib_timing: remove dependency from MARS and BRICK 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer f069de7226 infra: don't depend on build date/time 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 77a71012ae if: prefer bio_flagged() 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 81458e1467 proc: remove NET_COMPAT from upstream 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer ae01a3ac92 net: remove NET_COMPAT from upstream 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer e890b94741 net: remove CONFIG_MARS_NET_COMPAT from upstream 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 101c1759f4 infra: remove __HAS_NEW_FILLDIR_T from upstream 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer efe800857a light: remove BUILDTAG from upstream 2015-11-18 14:36:30 +01:00
Thomas Schoebel-Theuer 61377b468c if: rename MARS_MAJOR 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 5c014d2df5 make: substitute aio bricks with sio for kernel upstream 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 337738533f proc: substitute aio bricks with sio for kernel upstream 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 874ed253ef light: substitute aio bricks with sio for kernel upstream 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 30439fb85e server: substitute aio bricks with sio for kernel upstream 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer cda9b14450 brick_mem: make checkpatch.pl happy 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 4ed387196c infra: fix Makefile for upstream 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 0990c8c176 all: keep TRAILING_SEMICOLON happy 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 11728ec480 net: make SSCANF_TO_KSTRTO happy 2015-11-18 14:36:29 +01:00
Thomas Schoebel-Theuer 9b39888cb2 proc: make SSCANF_TO_KSTRTO happy 2015-11-18 14:36:29 +01:00