Message-ID: <YfWrpoQIR9+NDGHl@latitude>
Date:   Sat, 29 Jan 2022 22:03:34 +0100
From:   Jonathan Neuschäfer <j.neuschaefer@....net>
To:     "Jason A. Donenfeld" <Jason@...c4.com>
Cc:     Andy Lutomirski <luto@...capital.net>,
        Jonathan Neuschäfer <j.neuschaefer@....net>,
        Theodore Ts'o <tytso@....edu>,
        LKML <linux-kernel@...r.kernel.org>,
        Sebastian Andrzej Siewior <bigeasy@...utronix.de>,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>, Will Deacon <will@...nel.org>,
        Waiman Long <longman@...hat.com>,
        Boqun Feng <boqun.feng@...il.com>, sultan@...neltoast.com,
        Andy Lutomirski <luto@...nel.org>
Subject: Re: [PATCH v2] random: remove batched entropy locking

Hi Jason,

On Fri, Jan 28, 2022 at 11:35:48PM +0100, Jason A. Donenfeld wrote:
> Rather than use spinlocks to protect batched entropy, we can instead
> disable interrupts locally, since we're dealing with per-cpu data, and
> manage resets with a basic generation counter. This should fix up the
> below splat that Jonathan received with a PROVE_RAW_LOCK_NESTING=y
> kernel.
> 
> Note that Sebastian has pointed out a few other areas where
> using spinlock_t in an IRQ context is potentially problematic for
> PREEMPT_RT. This patch handles one of those cases, and we'll hopefully
> have additional patches for other cases.
> 
> [    2.500000] [ BUG: Invalid wait context ]
[...]
> 
> Reported-by: Jonathan Neuschäfer <j.neuschaefer@....net>
> Link: https://lore.kernel.org/lkml/YfMa0QgsjCVdRAvJ@latitude/
> Fixes: b7d5dc21072c ("random: add a spinlock_t to struct batched_entropy")
> Cc: Andy Lutomirski <luto@...nel.org>
> Cc: Sebastian Andrzej Siewior <bigeasy@...utronix.de>
> Signed-off-by: Jason A. Donenfeld <Jason@...c4.com>
> ---
> Changes v1->v2:
> - We move from Andy's original patch, which was a bit racey, to using a
>   simple generation counter.
> 
>  drivers/char/random.c | 58 ++++++++++++++++++++-----------------------
>  1 file changed, 27 insertions(+), 31 deletions(-)

This patch doesn't seem to apply properly on 5.17-rc1:

$ git am rand2.mbox
Applying: random: remove batched entropy locking
error: patch failed: drivers/char/random.c:2057
error: drivers/char/random.c: patch does not apply
Patch failed at 0001 random: remove batched entropy locking
hint: Use 'git am --show-current-patch=diff' to see the failed patch
When you have resolved this problem, run "git am --continue".
If you prefer to skip this patch, run "git am --skip" instead.
To restore the original branch and stop patching, run "git am --abort".

So I cherry-picked it from https://git.zx2c4.com/linux-rng jd/no-batch-lock.

The result looks alright (no splat during boot).

Tested-by: Jonathan Neuschäfer <j.neuschaefer@....net>
--


With both patches, I get the following about three minutes after boot, but
I guess it will be addressed by one of the other patchsets under discussion:

[  202.670000] =============================
[  202.670000] [ BUG: Invalid wait context ]
[  202.670000] 5.17.0-rc1-00001-g4e53c88bd88e #585 Not tainted
[  202.670000] -----------------------------
[  202.670000] swapper/0 is trying to lock:
[  202.670000] c0b0ff3c (random_write_wait.lock){....}-{3:3}, at: __wake_up_common_lock+0x54/0xb4
[  202.670000] other info that might help us debug this:
[  202.670000] context-{2:2}
[  202.670000] no locks held by swapper/0.
[  202.670000] stack backtrace:
[  202.670000] CPU: 0 PID: 0 Comm: swapper Not tainted 5.17.0-rc1-00001-g4e53c88bd88e #585
[  202.670000] Hardware name: WPCM450 chip
[  202.670000] [<c00100a8>] (unwind_backtrace) from [<c000db2c>] (show_stack+0x10/0x14)
[  202.670000] [<c000db2c>] (show_stack) from [<c0054eec>] (__lock_acquire+0x3f0/0x189c)
[  202.670000] [<c0054eec>] (__lock_acquire) from [<c0054478>] (lock_acquire+0x2b8/0x354)
[  202.670000] [<c0054478>] (lock_acquire) from [<c0568088>] (_raw_spin_lock_irqsave+0x60/0x74)
[  202.670000] [<c0568088>] (_raw_spin_lock_irqsave) from [<c004c34c>] (__wake_up_common_lock+0x54/0xb4)
[  202.670000] [<c004c34c>] (__wake_up_common_lock) from [<c004c3bc>] (__wake_up+0x10/0x18)
[  202.670000] [<c004c3bc>] (__wake_up) from [<c030d898>] (crng_reseed.constprop.0+0x240/0x338)
[  202.670000] [<c030d898>] (crng_reseed.constprop.0) from [<c0061b34>] (handle_irq_event_percpu+0x18/0x38)
[  202.670000] [<c0061b34>] (handle_irq_event_percpu) from [<c0061b8c>] (handle_irq_event+0x38/0x5c)
[  202.670000] [<c0061b8c>] (handle_irq_event) from [<c0065b28>] (handle_fasteoi_irq+0x9c/0x114)
[  202.670000] [<c0065b28>] (handle_fasteoi_irq) from [<c0061178>] (handle_irq_desc+0x24/0x34)
[  202.670000] [<c0061178>] (handle_irq_desc) from [<c05621ac>] (generic_handle_arch_irq+0x28/0x3c)
[  202.670000] [<c05621ac>] (generic_handle_arch_irq) from [<c0008eb4>] (__irq_svc+0x54/0x80)
[  202.670000] Exception stack(0xc0931f20 to 0xc0931f68)
[  202.670000] 1f20: 00000000 0005317f 0005217f 60000013 00000000 00000000 c0930000 ffffffff
[  202.670000] 1f40: c0937ac0 00000000 00053177 00000000 600000d3 c0931f70 c000b0e8 c000b0e0
[  202.670000] 1f60: 60000013 ffffffff
[  202.670000] [<c0008eb4>] (__irq_svc) from [<c000b0e0>] (arch_cpu_idle+0x24/0x34)
[  202.670000] [<c000b0e0>] (arch_cpu_idle) from [<c0567e24>] (default_idle_call+0x40/0x80)
[  202.670000] [<c0567e24>] (default_idle_call) from [<c0046fbc>] (do_idle+0xd4/0x254)
[  202.670000] [<c0046fbc>] (do_idle) from [<c00474e4>] (cpu_startup_entry+0xc/0x10)
[  202.670000] [<c00474e4>] (cpu_startup_entry) from [<c07bdedc>] (start_kernel+0x5cc/0x6c0)
[  202.670000] random: crng init done

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)

Powered by blists - more mailing lists