Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible
On Sep 12 2023, Liam R. Howlett wrote: > * Andreas Schwab [230912 14:15]: >> Any news? This is still broken. > > I have a proposed fix. I seem to have caused a pre-existing problem to > show up. Please see if the attached works for you, and I'll send it > to a lot of people. Thanks, it fixes the issue for me (tested both 6.5 and 6.6-rc1). -- Andreas Schwab, sch...@linux-m68k.org GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 "And now for something completely different."
Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible
* Andreas Schwab [230912 14:15]: > Any news? This is still broken. I have a proposed fix. I seem to have caused a pre-existing problem to show up. Please see if the attached works for you, and I'll send it to a lot of people. Thanks, Liam >From 9ef8f834bb0342dc26464b9dd0165929d3e6a7e5 Mon Sep 17 00:00:00 2001 From: "Liam R. Howlett" Date: Tue, 12 Sep 2023 13:45:29 -0400 Subject: [PATCH] init/main: Clear boot task idle flag Initial booting was setting the task flag to idle (PF_IDLE) by the call path sched_init() -> init_idle(). Having the task idle and calling call_rcu() in kernel/rcu/tiny.c means that TIF_NEED_RESCHED will be enabled. Subsequent calls to any cond_resched() will enable IRQs, potentially earlier than the enabling of IRQs. This causes a warning later in start_kernel() as interrupts are enabled before the are fully set up. Fix this issue by clearing the PF_IDLE flag on return from sched_init() and restore the flag in rest_init(). Signed-off-by: Liam R. Howlett --- init/main.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/init/main.c b/init/main.c index ad920fac325c..46b35be8f00a 100644 --- a/init/main.c +++ b/init/main.c @@ -696,7 +696,7 @@ noinline void __ref __noreturn rest_init(void) */ rcu_read_lock(); tsk = find_task_by_pid_ns(pid, _pid_ns); - tsk->flags |= PF_NO_SETAFFINITY; + tsk->flags |= PF_NO_SETAFFINITY | PF_IDLE; set_cpus_allowed_ptr(tsk, cpumask_of(smp_processor_id())); rcu_read_unlock(); @@ -938,6 +938,7 @@ void start_kernel(void) * time - but meanwhile we still have a functioning scheduler. */ sched_init(); + current->flags &= ~PF_IDLE; if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n")) -- 2.39.2
Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible
Any news? This is still broken. -- Andreas Schwab, sch...@linux-m68k.org GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 "And now for something completely different."
Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible
On Aug 31 2023, Michael Ellerman wrote: > Andreas Schwab writes: >> This breaks booting on ppc32: > > Does enabling CONFIG_DEBUG_ATOMIC_SLEEP fix the crash? Yes, it does. -- Andreas Schwab, sch...@linux-m68k.org GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 "And now for something completely different."
Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible
Andreas Schwab writes: > This breaks booting on ppc32: Does enabling CONFIG_DEBUG_ATOMIC_SLEEP fix the crash? It did for me on qemu. cheers > Kernel attemptd to writ user page (1ff0) - exploit attempt? (uid: 0) > BUG: Unable to handle kernel data access on write at 0x1ff0 > Faulting instruction address: 0xc0009554 > Vector: 300 (Data Access) at [c0b09d10] > pc: c0009554: do_softirq_own_stack+0x18/0x30 > lr: c004f480: __irq_exit_rcu+0x70/0xc0 > sp: c0b09dd0 >msr: 1032 >dar: 1ff0 > dsisr: 4200 > current = 0xc0a08360 > pid = 0, comm = swapper > Linux version 6.5.0 ... > enter ? for help > [c0b09de0] c00ff480 __irq_exit_rcu+0x70/0xc0 > [c0b09df0] c0005a98 Decrementer_virt+0x108/0x10c > --- Exception: 900 (Decrementer) at c06cfa0c __schedule+0x4fc/0x510 > [c0b09ec0] c06cf75c __schedule+0x1cc/0x510 (unreliable) > [c0b09ef0] c06cfc90 __cond_resched+0x2c/0x54 > [c0b09f00] c06d07f8 mutex_lock_killable+0x18/0x5c > [c0b09f10] c013c404 pcpu_alloc+0x110/0x4dc > [c0b09f70] c000cc34 alloc_descr.isra.18+0x48/0x144 > [c0b09f90] c0988aa0 early_irq_init+0x64/0x8c > [c0b09fa0] c097a5a4 start_kernel+0x5b4/0x7b0 > [c0b09ff0] 3dc0 > mon> > > -- > Andreas Schwab, sch...@linux-m68k.org > GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 > "And now for something completely different."
Re: [PATCH v2 1/2] maple_tree: Disable mas_wr_append() when other readers are possible
This breaks booting on ppc32: Kernel attemptd to writ user page (1ff0) - exploit attempt? (uid: 0) BUG: Unable to handle kernel data access on write at 0x1ff0 Faulting instruction address: 0xc0009554 Vector: 300 (Data Access) at [c0b09d10] pc: c0009554: do_softirq_own_stack+0x18/0x30 lr: c004f480: __irq_exit_rcu+0x70/0xc0 sp: c0b09dd0 msr: 1032 dar: 1ff0 dsisr: 4200 current = 0xc0a08360 pid = 0, comm = swapper Linux version 6.5.0 ... enter ? for help [c0b09de0] c00ff480 __irq_exit_rcu+0x70/0xc0 [c0b09df0] c0005a98 Decrementer_virt+0x108/0x10c --- Exception: 900 (Decrementer) at c06cfa0c __schedule+0x4fc/0x510 [c0b09ec0] c06cf75c __schedule+0x1cc/0x510 (unreliable) [c0b09ef0] c06cfc90 __cond_resched+0x2c/0x54 [c0b09f00] c06d07f8 mutex_lock_killable+0x18/0x5c [c0b09f10] c013c404 pcpu_alloc+0x110/0x4dc [c0b09f70] c000cc34 alloc_descr.isra.18+0x48/0x144 [c0b09f90] c0988aa0 early_irq_init+0x64/0x8c [c0b09fa0] c097a5a4 start_kernel+0x5b4/0x7b0 [c0b09ff0] 3dc0 mon> -- Andreas Schwab, sch...@linux-m68k.org GPG Key fingerprint = 7578 EB47 D4E5 4D69 2510 2552 DF73 E780 A9DA AEC1 "And now for something completely different."
