4.8 muqss #336

thunderrd · 2016-10-17T07:49:20Z

No description provided.

synchronise niffies to the highest value as idle ticks may have artificially kept niffies low on one CPU and the truth can only be later.

…changed.

…the current cpu.

…ies are altered by another runqueue.

…task functions to match core.c

improvements and logic fixes, including the ability to change iso tasks back to normal.

…topper task, removing the need for racy setting of task_cpu of it in set_task_cpu. Check against wake_cpu when calling set_task_cpu.

…ering correct current task under lock.

… or under lock

…lock

… to tasks bouncing to another CPU immediately instead of yielding leading to infinitely bouncing between CPUs. Improve behaviour by positively calling timeslice_expired.

…e choose the best policy as well as priority task from other runqueues. Check the value lockless first to avoid locking extra runqueues unless they're likely to contain a better choice.

A recent commit removed locking from netlink_diag_dump() but forgot one error case. ===================================== [ BUG: bad unlock balance detected! ] 4.9.0-rc3+ torvalds#336 Not tainted ------------------------------------- syz-executor/4018 is trying to release lock ([ 36.220068] nl_table_lock ) at: [<ffffffff82dc8683>] netlink_diag_dump+0x1a3/0x250 net/netlink/diag.c:182 but there are no more locks to release! other info that might help us debug this: 3 locks held by syz-executor/4018: #0: [ 36.220068] ( sock_diag_mutex[ 36.220068] ){+.+.+.} , at: [ 36.220068] [<ffffffff82c3873b>] sock_diag_rcv+0x1b/0x40 #1: [ 36.220068] ( sock_diag_table_mutex[ 36.220068] ){+.+.+.} , at: [ 36.220068] [<ffffffff82c38e00>] sock_diag_rcv_msg+0x140/0x3a0 #2: [ 36.220068] ( nlk->cb_mutex[ 36.220068] ){+.+.+.} , at: [ 36.220068] [<ffffffff82db6600>] netlink_dump+0x50/0xac0 stack backtrace: CPU: 1 PID: 4018 Comm: syz-executor Not tainted 4.9.0-rc3+ torvalds#336 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 ffff8800645df688 ffffffff81b46934 ffffffff84eb3e78 ffff88006ad85800 ffffffff82dc8683 ffffffff84eb3e78 ffff8800645df6b8 ffffffff812043ca dffffc0000000000 ffff88006ad85ff8 ffff88006ad85fd0 00000000ffffffff Call Trace: [< inline >] __dump_stack lib/dump_stack.c:15 [<ffffffff81b46934>] dump_stack+0xb3/0x10f lib/dump_stack.c:51 [<ffffffff812043ca>] print_unlock_imbalance_bug+0x17a/0x1a0 kernel/locking/lockdep.c:3388 [< inline >] __lock_release kernel/locking/lockdep.c:3512 [<ffffffff8120cfd8>] lock_release+0x8e8/0xc60 kernel/locking/lockdep.c:3765 [< inline >] __raw_read_unlock ./include/linux/rwlock_api_smp.h:225 [<ffffffff83fc001a>] _raw_read_unlock+0x1a/0x30 kernel/locking/spinlock.c:255 [<ffffffff82dc8683>] netlink_diag_dump+0x1a3/0x250 net/netlink/diag.c:182 [<ffffffff82db6947>] netlink_dump+0x397/0xac0 net/netlink/af_netlink.c:2110 Fixes: ad20207 ("netlink: Use rhashtable walk interface in diag dump") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Andrey Konovalov <andreyknvl@google.com> Tested-by: Andrey Konovalov <andreyknvl@google.com>

A recent commit removed locking from netlink_diag_dump() but forgot one error case. ===================================== [ BUG: bad unlock balance detected! ] 4.9.0-rc3+ torvalds#336 Not tainted ------------------------------------- syz-executor/4018 is trying to release lock ([ 36.220068] nl_table_lock ) at: [<ffffffff82dc8683>] netlink_diag_dump+0x1a3/0x250 net/netlink/diag.c:182 but there are no more locks to release! other info that might help us debug this: 3 locks held by syz-executor/4018: #0: [ 36.220068] ( sock_diag_mutex[ 36.220068] ){+.+.+.} , at: [ 36.220068] [<ffffffff82c3873b>] sock_diag_rcv+0x1b/0x40 #1: [ 36.220068] ( sock_diag_table_mutex[ 36.220068] ){+.+.+.} , at: [ 36.220068] [<ffffffff82c38e00>] sock_diag_rcv_msg+0x140/0x3a0 #2: [ 36.220068] ( nlk->cb_mutex[ 36.220068] ){+.+.+.} , at: [ 36.220068] [<ffffffff82db6600>] netlink_dump+0x50/0xac0 stack backtrace: CPU: 1 PID: 4018 Comm: syz-executor Not tainted 4.9.0-rc3+ torvalds#336 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 ffff8800645df688 ffffffff81b46934 ffffffff84eb3e78 ffff88006ad85800 ffffffff82dc8683 ffffffff84eb3e78 ffff8800645df6b8 ffffffff812043ca dffffc0000000000 ffff88006ad85ff8 ffff88006ad85fd0 00000000ffffffff Call Trace: [< inline >] __dump_stack lib/dump_stack.c:15 [<ffffffff81b46934>] dump_stack+0xb3/0x10f lib/dump_stack.c:51 [<ffffffff812043ca>] print_unlock_imbalance_bug+0x17a/0x1a0 kernel/locking/lockdep.c:3388 [< inline >] __lock_release kernel/locking/lockdep.c:3512 [<ffffffff8120cfd8>] lock_release+0x8e8/0xc60 kernel/locking/lockdep.c:3765 [< inline >] __raw_read_unlock ./include/linux/rwlock_api_smp.h:225 [<ffffffff83fc001a>] _raw_read_unlock+0x1a/0x30 kernel/locking/spinlock.c:255 [<ffffffff82dc8683>] netlink_diag_dump+0x1a3/0x250 net/netlink/diag.c:182 [<ffffffff82db6947>] netlink_dump+0x397/0xac0 net/netlink/af_netlink.c:2110 Fixes: ad20207 ("netlink: Use rhashtable walk interface in diag dump") Signed-off-by: Eric Dumazet <edumazet@google.com> Reported-by: Andrey Konovalov <andreyknvl@google.com> Tested-by: Andrey Konovalov <andreyknvl@google.com> Signed-off-by: David S. Miller <davem@davemloft.net>

switch to idle_host_task when idle

CI: build kernel with clang/llvm 12

The current source pushes skb into dev->done queue by calling skb_queue_tail() and then, call skb_dequeue() to pop for rx_cleanup state to free urb and skb next in usbnet_bh(). It wastes CPU resource with extra instructions. Instead, use return values jumping to rx_cleanup case directly to free them. Therefore calling skb_queue_tail() and skb_dequeue() is not necessary. The follows are just showing difference between calling skb_queue_tail() and using return values jumping to rx_cleanup state directly in usbnet_bh() in Arm64 instructions with perf tool. ----------- calling skb_queue_tail() ----------- │ if (!(dev->driver_info->flags & FLAG_RX_ASSEMBLE)) 7.58 │248: ldr x0, [x20, torvalds#16] 2.46 │24c: ldr w0, [x0, torvalds#8] 1.64 │250: ↑ tbnz w0, torvalds#14, 16c │ dev->net->stats.rx_errors++; 0.57 │254: ldr x1, [x20, torvalds#184] 1.64 │258: ldr x0, [x1, torvalds#336] 2.65 │25c: add x0, x0, #0x1 │260: str x0, [x1, torvalds#336] │ skb_queue_tail(&dev->done, skb); 0.38 │264: mov x1, x19 │268: mov x0, x21 2.27 │26c: → bl skb_queue_tail 0.57 │270: ↑ b 44 // branch to call skb_dequeue() ----------- jumping to rx_cleanup state ----------- │ if (!(dev->driver_info->flags & FLAG_RX_ASSEMBLE)) 1.69 │25c: ldr x0, [x21, torvalds#16] 4.78 │260: ldr w0, [x0, torvalds#8] 3.28 │264: ↑ tbnz w0, torvalds#14, e4 // jump to 'rx_cleanup' state │ dev->net->stats.rx_errors++; 0.09 │268: ldr x1, [x21, torvalds#184] 2.72 │26c: ldr x0, [x1, torvalds#336] 3.37 │270: add x0, x0, #0x1 0.09 │274: str x0, [x1, torvalds#336] 0.66 │278: ↑ b e4 // branch to 'rx_cleanup' state Signed-off-by: Leesoo Ahn <lsahn@ooseel.net>

Add a test case which replaces an active ingress qdisc while keeping the miniq in-tact during the transition period to the new clsact qdisc. # ./vmtest.sh -- ./test_progs -t tc_link [...] ./test_progs -t tc_link [ 3.412871] bpf_testmod: loading out-of-tree module taints kernel. [ 3.413343] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel torvalds#332 tc_links_after:OK torvalds#333 tc_links_append:OK torvalds#334 tc_links_basic:OK torvalds#335 tc_links_before:OK torvalds#336 tc_links_chain_classic:OK torvalds#337 tc_links_chain_mixed:OK torvalds#338 tc_links_dev_chain0:OK torvalds#339 tc_links_dev_cleanup:OK torvalds#340 tc_links_dev_mixed:OK torvalds#341 tc_links_ingress:OK torvalds#342 tc_links_invalid:OK torvalds#343 tc_links_prepend:OK torvalds#344 tc_links_replace:OK torvalds#345 tc_links_revision:OK Summary: 14/0 PASSED, 0 SKIPPED, 0 FAILED Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Martin KaFai Lau <martin.lau@kernel.org> Link: https://lore.kernel.org/r/20240708133130.11609-2-daniel@iogearbox.net Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>

[ Upstream commit 5f1d18d ] Add a test case which replaces an active ingress qdisc while keeping the miniq in-tact during the transition period to the new clsact qdisc. # ./vmtest.sh -- ./test_progs -t tc_link [...] ./test_progs -t tc_link [ 3.412871] bpf_testmod: loading out-of-tree module taints kernel. [ 3.413343] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel torvalds#332 tc_links_after:OK torvalds#333 tc_links_append:OK torvalds#334 tc_links_basic:OK torvalds#335 tc_links_before:OK torvalds#336 tc_links_chain_classic:OK torvalds#337 tc_links_chain_mixed:OK torvalds#338 tc_links_dev_chain0:OK torvalds#339 tc_links_dev_cleanup:OK torvalds#340 tc_links_dev_mixed:OK torvalds#341 tc_links_ingress:OK torvalds#342 tc_links_invalid:OK torvalds#343 tc_links_prepend:OK torvalds#344 tc_links_replace:OK torvalds#345 tc_links_revision:OK Summary: 14/0 PASSED, 0 SKIPPED, 0 FAILED Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Martin KaFai Lau <martin.lau@kernel.org> Link: https://lore.kernel.org/r/20240708133130.11609-2-daniel@iogearbox.net Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org> Signed-off-by: Sasha Levin <sashal@kernel.org>

Add a test case which replaces an active ingress qdisc while keeping the miniq in-tact during the transition period to the new clsact qdisc. # ./vmtest.sh -- ./test_progs -t tc_link [...] ./test_progs -t tc_link [ 3.412871] bpf_testmod: loading out-of-tree module taints kernel. [ 3.413343] bpf_testmod: module verification failed: signature and/or required key missing - tainting kernel torvalds#332 tc_links_after:OK torvalds#333 tc_links_append:OK torvalds#334 tc_links_basic:OK torvalds#335 tc_links_before:OK torvalds#336 tc_links_chain_classic:OK torvalds#337 tc_links_chain_mixed:OK torvalds#338 tc_links_dev_chain0:OK torvalds#339 tc_links_dev_cleanup:OK torvalds#340 tc_links_dev_mixed:OK torvalds#341 tc_links_ingress:OK torvalds#342 tc_links_invalid:OK torvalds#343 tc_links_prepend:OK torvalds#344 tc_links_replace:OK torvalds#345 tc_links_revision:OK Summary: 14/0 PASSED, 0 SKIPPED, 0 FAILED Signed-off-by: Daniel Borkmann <daniel@iogearbox.net> Cc: Martin KaFai Lau <martin.lau@kernel.org> Link: https://lore.kernel.org/r/20240708133130.11609-2-daniel@iogearbox.net Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>

ckolivas added 30 commits October 9, 2016 17:46

MuQSS version 0.104

36acb02

MuQSS version 0.105

24ba405

MuQSS version 0.106

560a7c9

MuQSS version 0.108

386dd43

muqss108-001-check_affinity_switch

659a4a8

muqss108-002-bias_idle_on_wake

1641360

Any time we have two runqueues locked we use that as an opportunity to

225a663

synchronise niffies to the highest value as idle ticks may have artificially kept niffies low on one CPU and the truth can only be later.

Don't reinitialise deadline on wake up new task in case runqueue has …

3201c4e

…changed.

sched_info_de/queued only on de/activate.

8d5ded8

Lock pi_lock as well when migrating a task in finish_lock_switch.

31a3fa8

Drop task waking which is not meaningfully used.

b1068ce

Do not update task_thread_info cpu on a running task until it is off …

b9889eb

…the current cpu.

Consolidate when and where we update_clocks and account for when niff…

4c1b1f5

…ies are altered by another runqueue.

Update sched_info only on first enqueue/dequeue and alter en/dequeue_…

3688005

…task functions to match core.c

Wrong again

f12d07e

Should update sched info data when moving CPUs to get correct rq clock.

0f6b426

Rework SCHED_ISO to work per-runqueue with numerous updates, overhead

182a0ce

improvements and logic fixes, including the ability to change iso tasks back to normal.

Build fixes for various different stripped configurations.

b8e3f69

Revert unnecessary moving of preempt_enable.

7d2ed0b

Bump MuQSS version to 0.110

73e3f98

Select a valid CPU from the online masks only

753f3af

Fix suspend and resume.

a25a303

Bump MuQSS version to 0.111

2c80363

Clean up bind_zero to not try and change affinity or reschedule the s…

fdd879d

…topper task, removing the need for racy setting of task_cpu of it in set_task_cpu. Check against wake_cpu when calling set_task_cpu.

Remove rq_policy which is only used locally

74ccb93

Double lock and differentiate rq from new_rq in wake_up_new_task, alt…

121b4e9

…ering correct current task under lock.

Remove use of rq_time_slice as time_slice is now only checked locally…

fb66efa

… or under lock

Remove use of rq_last_ran as it is now only checked locally or under …

e34f461

…lock

Update sched info when we change CPUs and not on en/dequeuing to be run

83d4c55

Remove unused code.

73a10ed

ckolivas added 13 commits October 15, 2016 09:25

Avoid spurious missed preemption due to lockless resched_curr.

1a62697

Time slice expire should be set to correct rq

7e45f4a

Remove redundant reassignment

ab70a89

Use rq dither as an offset itself, avoiding one conditional

9989220

Dequeue sched info on task deactivation

0709a96

Do irq_enter on scheduler_ipi called when idle to update xtime.

402bc62

Don't re-set values that haven't changed.

755fcbd

Yielding repeatedly without resetting timeslice and deadline can lead…

2ce037a

… to tasks bouncing to another CPU immediately instead of yielding leading to infinitely bouncing between CPUs. Improve behaviour by positively calling timeslice_expired.

Remove dup'd vtime_task_switch

3e4e282

Add more documentation to sched-MuQSS.txt.

046c8df

Choose deadline task by skip list key instead of deadline to ensure w…

a0a5f7e

…e choose the best policy as well as priority task from other runqueues. Check the value lockless first to avoid locking extra runqueues unless they're likely to contain a better choice.

Bump MuQSS version to 0.112

2ddc12e

Typo

75f10f0

thunderrd closed this Oct 17, 2016

Mic92 pushed a commit to Mic92/linux that referenced this pull request Feb 4, 2019

Merge pull request torvalds#336 from laijs/lkl/idle_host_task

be0ea3f

switch to idle_host_task when idle

ojeda added a commit to ojeda/linux that referenced this pull request Jun 3, 2021

Merge pull request torvalds#336 from TheSven73/rust-for-linux-clang-12

77316f7

CI: build kernel with clang/llvm 12

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

4.8 muqss #336

4.8 muqss #336

thunderrd commented Oct 17, 2016

4.8 muqss #336

4.8 muqss #336

Conversation

thunderrd commented Oct 17, 2016