PF_IO_WORKER are kernel threads too, but they aren't PF_KTHREAD in the sense that we don't assign ->set_child_tid with our own structure. Just ensure that every arch sets up the PF_IO_WORKER threads like kthreads in the arch implementation of copy_thread(). Signed-off-by: Jens Axboe <axboe@kernel.dk>
Linux 5.11.rcX was failing to boot on ARC HSDK board. Turns out we have
a couple of issues, this being the first one, and I'm to blame as I
didn't pay attention during review.
TIF_NOTIFY_SIGNAL support requires checking multiple TIF_* bits in
kernel return code path. Old code only needed to check a single bit so
BBIT0 <TIF_SIGPENDING> worked. New code needs to check multiple bits so
AND <bit-mask> instruction. So needs to use bit mask variant _TIF_SIGPENDING
Cc: Jens Axboe <axboe@kernel.dk>
Fixes: 53855e12588743ea128 ("arc: add support for TIF_NOTIFY_SIGNAL")
Link: https://github.com/foss-for-synopsys-dwc-arc-processors/linux/issues/34
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Pull TIF_NOTIFY_SIGNAL updates from Jens Axboe: "This sits on top of of the core entry/exit and x86 entry branch from the tip tree, which contains the generic and x86 parts of this work. Here we convert the rest of the archs to support TIF_NOTIFY_SIGNAL. With that done, we can get rid of JOBCTL_TASK_WORK from task_work and signal.c, and also remove a deadlock work-around in io_uring around knowing that signal based task_work waking is invoked with the sighand wait queue head lock. The motivation for this work is to decouple signal notify based task_work, of which io_uring is a heavy user of, from sighand. The sighand lock becomes a huge contention point, particularly for threaded workloads where it's shared between threads. Even outside of threaded applications it's slower than it needs to be. Roman Gershman <romger@amazon.com> reported that his networked workload dropped from 1.6M QPS at 80% CPU to 1.0M QPS at 100% CPU after io_uring was changed to use TIF_NOTIFY_SIGNAL. The time was all spent hammering on the sighand lock, showing 57% of the CPU time there [1]. There are further cleanups possible on top of this. One example is TIF_PATCH_PENDING, where a patch already exists to use TIF_NOTIFY_SIGNAL instead. Hopefully this will also lead to more consolidation, but the work stands on its own as well" [1] https://github.com/axboe/liburing/issues/215 * tag 'tif-task_work.arch-2020-12-14' of git://git.kernel.dk/linux-block: (28 commits) io_uring: remove 'twa_signal_ok' deadlock work-around kernel: remove checking for TIF_NOTIFY_SIGNAL signal: kill JOBCTL_TASK_WORK io_uring: JOBCTL_TASK_WORK is no longer used by task_work task_work: remove legacy TWA_SIGNAL path sparc: add support for TIF_NOTIFY_SIGNAL riscv: add support for TIF_NOTIFY_SIGNAL nds32: add support for TIF_NOTIFY_SIGNAL ia64: add support for TIF_NOTIFY_SIGNAL h8300: add support for TIF_NOTIFY_SIGNAL c6x: add support for TIF_NOTIFY_SIGNAL alpha: add support for TIF_NOTIFY_SIGNAL xtensa: add support for TIF_NOTIFY_SIGNAL arm: add support for TIF_NOTIFY_SIGNAL microblaze: add support for TIF_NOTIFY_SIGNAL hexagon: add support for TIF_NOTIFY_SIGNAL csky: add support for TIF_NOTIFY_SIGNAL openrisc: add support for TIF_NOTIFY_SIGNAL sh: add support for TIF_NOTIFY_SIGNAL um: add support for TIF_NOTIFY_SIGNAL ...
This is a non-functional change, if anything a better fall-back handling. Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
To start stack unwinding (SP, PC and BLINK) are needed. When the explicit execution context (pt_regs etc) is not available, unwinder assumes the task is sleeping (in __switch_to()) and fetches SP and BLINK from kernel mode stack. But this assumption is not true, specially in a SMP system, when top runs on 1 core, there may be active running processes on all cores. So when unwinding non courrent tasks, ensure they are NOT running. And while at it, handle the self unwinding case explicitly. This came out of investigation of a customer reported hang with rcutorture+top Link: https://github.com/foss-for-synopsys-dwc-arc-processors/linux/issues/31 Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Wire up TIF_NOTIFY_SIGNAL handling for arc. Cc: linux-snps-arc@lists.infradead.org Acked-by: Vineet Gupta <vgupta@synopsys.com> Signed-off-by: Jens Axboe <axboe@kernel.dk>
ARC HSDK platform stopped booting on released v5.10-rc1, getting stuck in startup of non master SMP cores. This was bisected to upstream commit 7fef431be9c9ac25 "(mm/page_alloc: place pages to tail in __free_pages_core())" That commit itself is harmless, it just exposed a subtle assumption in our platform code (hence CC'ing linux-mm just as FYI in case some other arches / platforms trip on it). The upstream commit is semantically disruptive as it reverses the order of page allocations (actually it can be good test for hardware verification to exercise different memory patterns altogether). For ARC HSDK platform that meant a remapped memory region (pertaining to unused Closely Coupled Memory) started getting used early for dynamice allocations, while not effectively remapped on all the cores, triggering memory error exception on those cores. The fix is to move the CCM remapping from early platform code to to early core boot code. And while it is undesirable to riddle common boot code with platform quirks, there is no other way to do this since the faltering code involves setting up stack itself so even function calls are not allowed at that point. If anyone is interested, all the gory details can be found at Link below. Link: https://github.com/foss-for-synopsys-dwc-arc-processors/linux/issues/32 Cc: David Hildenbrand <david@redhat.com> Cc: linux-mm@kvack.org Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Currently stack unwinder is a while(1) loop which relies on the dwarf unwinder to signal termination, which in turn relies on dwarf info to do so. This in theory could cause an infinite loop if the dwarf info was somehow messed up or the register contents were etc. This fix thus detects the excessive looping and breaks the loop. | Mem: 26184K used, 1009136K free, 0K shrd, 0K buff, 14416K cached | CPU: 0.0% usr 72.8% sys 0.0% nic 27.1% idle 0.0% io 0.0% irq 0.0% sirq | Load average: 4.33 2.60 1.11 2/74 139 | PID PPID USER STAT VSZ %VSZ CPU %CPU COMMAND | 133 2 root SWN 0 0.0 3 22.9 [rcu_torture_rea] | 132 2 root SWN 0 0.0 0 22.0 [rcu_torture_rea] | 131 2 root SWN 0 0.0 3 21.5 [rcu_torture_rea] | 126 2 root RW 0 0.0 2 5.4 [rcu_torture_wri] | 129 2 root SWN 0 0.0 0 0.2 [rcu_torture_fak] | 137 2 root SW 0 0.0 0 0.2 [rcu_torture_cbf] | 127 2 root SWN 0 0.0 0 0.1 [rcu_torture_fak] | 138 115 root R 1464 0.1 2 0.1 top | 130 2 root SWN 0 0.0 0 0.1 [rcu_torture_fak] | 128 2 root SWN 0 0.0 0 0.1 [rcu_torture_fak] | 115 1 root S 1472 0.1 1 0.0 -/bin/sh | 104 1 root S 1464 0.1 0 0.0 inetd | 1 0 root S 1456 0.1 2 0.0 init | 78 1 root S 1456 0.1 0 0.0 syslogd -O /var/log/messages | 134 2 root SW 0 0.0 2 0.0 [rcu_torture_sta] | 10 2 root IW 0 0.0 1 0.0 [rcu_preempt] | 88 2 root IW 0 0.0 1 0.0 [kworker/1:1-eve] | 66 2 root IW 0 0.0 2 0.0 [kworker/2:2-eve] | 39 2 root IW 0 0.0 2 0.0 [kworker/2:1-eve] | unwinder looping too long, aborting ! Cc: <stable@vger.kernel.org> Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Pull arch task_work cleanups from Jens Axboe:
"Two cleanups that don't fit other categories:
- Finally get the task_work_add() cleanup done properly, so we don't
have random 0/1/false/true/TWA_SIGNAL confusing use cases. Updates
all callers, and also fixes up the documentation for
task_work_add().
- While working on some TIF related changes for 5.11, this
TIF_NOTIFY_RESUME cleanup fell out of that. Remove some arch
duplication for how that is handled"
* tag 'arch-cleanup-2020-10-22' of git://git.kernel.dk/linux-block:
task_work: cleanup notification modes
tracehook: clear TIF_NOTIFY_RESUME in tracehook_notify_resume()
Pull ARC fix from Vineet Gupta: "I found a snafu in perf driver which made it into 5.9-rc4 and the fix should go in now than wait" * tag 'arc-5.10-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc: ARC: perf: redo the pct irq missing in device-tree handling