sci_request_irq() checks port->irqstr[j] for a NULL pointer, to decide
if a fallback interrupt name string should be allocated or not.
While this string is freed during port shutdown, the pointer is not
zeroed. Hence on a subsequent startup of the port, it will still be
pointing to the freed memory, leading to e.g.
WARNING: CPU: 0 PID: 404 at fs/proc/generic.c:388 __proc_create+0xbc/0x260
name len 0
or to a crash (the latter is more likely with CONFIG_DEBUG_SLAB=y, due
to the poisoning of freed memory).
Instead of zeroeing the pointer at multiple places, preinitialize
port->irqstr[j] to zero to fix this.
Fixes: 8b0bbd956228ae87 ("serial: sh-sci: Add support for R7S9210")
Signed-off-by: Geert Uytterhoeven <[email protected]>
---
drivers/tty/serial/sh-sci.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
index 5808edfe3f7be404..f8e53ac5c17dfb94 100644
--- a/drivers/tty/serial/sh-sci.c
+++ b/drivers/tty/serial/sh-sci.c
@@ -1872,6 +1872,7 @@ static int sci_request_irq(struct sci_port *port)
}
desc = sci_irq_desc + i;
+ port->irqstr[j] = NULL;
if (SCIx_TEIDRI_IRQ_EXISTS(port)) {
/*
* ERI and BRI are muxed, just register ERI and
--
2.17.1
On Tue, Jul 17, 2018 at 10:58 AM, Geert Uytterhoeven
<[email protected]> wrote:
> sci_request_irq() checks port->irqstr[j] for a NULL pointer, to decide
> if a fallback interrupt name string should be allocated or not.
>
> While this string is freed during port shutdown, the pointer is not
> zeroed. Hence on a subsequent startup of the port, it will still be
> pointing to the freed memory, leading to e.g.
>
> WARNING: CPU: 0 PID: 404 at fs/proc/generic.c:388 __proc_create+0xbc/0x260
> name len 0
>
> or to a crash (the latter is more likely with CONFIG_DEBUG_SLAB=y, due
> to the poisoning of freed memory).
>
> Instead of zeroeing the pointer at multiple places, preinitialize
> port->irqstr[j] to zero to fix this.
>
> Fixes: 8b0bbd956228ae87 ("serial: sh-sci: Add support for R7S9210")
> Signed-off-by: Geert Uytterhoeven <[email protected]>
> ---
> drivers/tty/serial/sh-sci.c | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/drivers/tty/serial/sh-sci.c b/drivers/tty/serial/sh-sci.c
> index 5808edfe3f7be404..f8e53ac5c17dfb94 100644
> --- a/drivers/tty/serial/sh-sci.c
> +++ b/drivers/tty/serial/sh-sci.c
> @@ -1872,6 +1872,7 @@ static int sci_request_irq(struct sci_port *port)
> }
>
> desc = sci_irq_desc + i;
> + port->irqstr[j] = NULL;
> if (SCIx_TEIDRI_IRQ_EXISTS(port)) {
> /*
> * ERI and BRI are muxed, just register ERI and
> --
> 2.17.1
>
Reviewed-by: Ulrich Hecht <[email protected]>
CU
Uli
On Tue, Jul 17, 2018 at 10:58:10AM +0200, Geert Uytterhoeven wrote:
> sci_request_irq() checks port->irqstr[j] for a NULL pointer, to decide
> if a fallback interrupt name string should be allocated or not.
>
> While this string is freed during port shutdown, the pointer is not
> zeroed. Hence on a subsequent startup of the port, it will still be
> pointing to the freed memory, leading to e.g.
>
> WARNING: CPU: 0 PID: 404 at fs/proc/generic.c:388 __proc_create+0xbc/0x260
> name len 0
>
> or to a crash (the latter is more likely with CONFIG_DEBUG_SLAB=y, due
> to the poisoning of freed memory).
>
> Instead of zeroeing the pointer at multiple places, preinitialize
> port->irqstr[j] to zero to fix this.
>
> Fixes: 8b0bbd956228ae87 ("serial: sh-sci: Add support for R7S9210")
> Signed-off-by: Geert Uytterhoeven <[email protected]>
Reviewed-by: Simon Horman <[email protected]>