Syzkaller reports memory leak issue at ieee80211_ibss_rx_no_sta()
in 5.10 stable releases. The problem has been fixed by the following patch
which can be cleanly applied to the 5.10 branch.
Found by Linux Verification Center (linuxtesting.org) with Syzkaller.
Hi all,
I wonder if there was an attempt to backport d4ae9916ea29 ("mm: soft-offline:
close the race against page allocation") from 4.19 to 4.14 and 4.9 (patch does
not apply, haven't found anything on stable ML, nor in stable tree git,
therefore I assume it was left as not easily fixable).
I'm asking because William is writing a LTP test madvise11 [1] which shows it's
failing on 4.14.290 (the latest 4.14). I know that 4.9 EOL in 4 months, but 4.14
in Jan 2024, it might be worth to fix it at least for 4.14.
Kind regards,
Petr
[1] https://lore.kernel.org/ltp/1659975072-29808-1-git-send-email-william.roche…
On Sat, 13 Aug 2022, Sasha Levin wrote:
> This is a note to let you know that I've just added the patch titled
>
> serial: Store character timing information to uart_port
>
> to the 5.18-stable tree which can be found at:
> http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum…
>
> The filename of the patch is:
> serial-store-character-timing-information-to-uart_po.patch
> and it can be found in the queue-5.18 subdirectory.
>
> If you, or anyone else, feels it should not be added to the stable tree,
> please let <stable(a)vger.kernel.org> know about it.
>
> commit 5517053a2e0b30a1e35f90504446af4a2c4920e8
> Author: Ilpo Järvinen <ilpo.jarvinen(a)linux.intel.com>
> Date: Mon Apr 25 17:33:58 2022 +0300
>
> serial: Store character timing information to uart_port
>
> [ Upstream commit 31f6bd7fad3b149a1eb6f67fc2e742e4df369b3d ]
>
> Struct uart_port currently stores FIFO timeout. Having character timing
> information readily available is useful. Even serial core itself
> determines char_time from port->timeout using inverse calculation.
>
> Store frame_time directly into uart_port. Character time is stored in
> nanoseconds to have reasonable precision with high rates. To avoid
> overflow, 64-bit math is necessary.
>
> It might be possible to determine timeout from frame_time by
> multiplying it with fifosize as needed but only part of the users seem
> to be protected by a lock. Thus, this patch does not pursue storing
> only frame_time in uart_port.
>
> Signed-off-by: Ilpo Järvinen <ilpo.jarvinen(a)linux.intel.com>
> Link: https://lore.kernel.org/r/20220425143410.12703-2-ilpo.jarvinen@linux.intel.…
> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org>
> Signed-off-by: Sasha Levin <sashal(a)kernel.org>
>
> diff --git a/drivers/tty/serial/serial_core.c b/drivers/tty/serial/serial_core.c
> index 95d8d1fcd543..6b07b7b41354 100644
> --- a/drivers/tty/serial/serial_core.c
> +++ b/drivers/tty/serial/serial_core.c
> @@ -24,6 +24,7 @@
> #include <linux/sysrq.h>
> #include <linux/delay.h>
> #include <linux/mutex.h>
> +#include <linux/math64.h>
> #include <linux/security.h>
>
> #include <linux/irq.h>
> @@ -333,15 +334,18 @@ void
> uart_update_timeout(struct uart_port *port, unsigned int cflag,
> unsigned int baud)
> {
> - unsigned int size;
> + unsigned int size = tty_get_frame_size(cflag);
> + u64 frame_time;
>
> - size = tty_get_frame_size(cflag) * port->fifosize;
> + frame_time = (u64)size * NSEC_PER_SEC;
> + size *= port->fifosize;
>
> /*
> * Figure the timeout to send the above number of bits.
> * Add .02 seconds of slop
> */
> port->timeout = (HZ * size) / baud + HZ/50;
> + port->frame_time = DIV64_U64_ROUND_UP(frame_time, baud);
> }
> EXPORT_SYMBOL(uart_update_timeout);
>
> @@ -1610,10 +1614,8 @@ static void uart_wait_until_sent(struct tty_struct *tty, int timeout)
> * Note: we have to use pretty tight timings here to satisfy
> * the NIST-PCTS.
> */
> - char_time = (port->timeout - HZ/50) / port->fifosize;
> - char_time = char_time / 5;
> - if (char_time == 0)
> - char_time = 1;
> + char_time = max(nsecs_to_jiffies(port->frame_time / 5), 1UL);
> +
> if (timeout && timeout < char_time)
> char_time = timeout;
>
> diff --git a/include/linux/serial_core.h b/include/linux/serial_core.h
> index ca57d686d4d1..409573ea5ea4 100644
> --- a/include/linux/serial_core.h
> +++ b/include/linux/serial_core.h
> @@ -232,6 +232,7 @@ struct uart_port {
> int hw_stopped; /* sw-assisted CTS flow state */
> unsigned int mctrl; /* current modem ctrl settings */
> unsigned int timeout; /* character-based timeout */
> + unsigned int frame_time; /* frame timing in ns */
> unsigned int type; /* port type */
> const struct uart_ops *ops;
> unsigned int custom_divisor;
>
Why is this change necessary for stable? The change was made simply to
make some other feature changes following a lot simpler but I don't think
there is any known bug/issue it would be fixing.
--
i.
This patch replaces some error-prone ~350 line long C&P initialization
code with a simple for-loop doing the same.
Conflict resolution was required due to commits missing in 4.9:
- ca97d939 (security: mark LSM hooks as __ro_after_init)
- 791ec491 (prlimit,security,selinux: add a security hook for prlimit)
- d69dece5 (LSM: Add /sys/kernel/security/lsm)
Those add new (security) features and hence are now allowed according to
the stable kernel rules. See e.g. rejection of the first patch for 4.9 in
https://lore.kernel.org/all/YsrfDfe3urGkepvJ@kroah.com/
If any of the above commits are reconsidered for 4.9 I can provide
backports of those.
Tetsuo Handa (1):
LSM: Initialize security_hook_heads upon registration.
security/security.c | 359 +-------------------------------------------
1 file changed, 7 insertions(+), 352 deletions(-)
--
2.25.1