linux-i2c.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] i2c: i801: Don't generate an interrupt on bus reset
@ 2021-05-25 15:03 Jean Delvare
  2021-05-26  9:10 ` Jarkko Nikula
  2021-05-27 19:56 ` Wolfram Sang
  0 siblings, 2 replies; 3+ messages in thread
From: Jean Delvare @ 2021-05-25 15:03 UTC (permalink / raw)
  To: Linux I2C; +Cc: Andy Shevchenko, Jarkko Nikula, Heiner Kallweit

Now that the i2c-i801 driver supports interrupts, setting the KILL bit
in a attempt to recover from a timed out transaction triggers an
interrupt. Unfortunately, the interrupt handler (i801_isr) is not
prepared for this situation and will try to process the interrupt as
if it was signaling the end of a successful transaction. In the case
of a block transaction, this can result in an out-of-range memory
access.

This condition was reproduced several times by syzbot:
https://syzkaller.appspot.com/bug?extid=ed71512d469895b5b34e
https://syzkaller.appspot.com/bug?extid=8c8dedc0ba9e03f6c79e
https://syzkaller.appspot.com/bug?extid=c8ff0b6d6c73d81b610e
https://syzkaller.appspot.com/bug?extid=33f6c360821c399d69eb
https://syzkaller.appspot.com/bug?extid=be15dc0b1933f04b043a
https://syzkaller.appspot.com/bug?extid=b4d3fd1dfd53e90afd79

So disable interrupts while trying to reset the bus. Interrupts will
be enabled again for the following transaction.

Fixes: 636752bcb517 ("i2c-i801: Enable IRQ for SMBus transactions")
Reported-by: syzbot+b4d3fd1dfd53e90afd79@syzkaller.appspotmail.com
Signed-off-by: Jean Delvare <jdelvare@suse.de>
Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Jarkko Nikula <jarkko.nikula@linux.intel.com>
---
 drivers/i2c/busses/i2c-i801.c |    6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

--- linux-5.12.orig/drivers/i2c/busses/i2c-i801.c	2021-05-24 12:00:59.307576983 +0200
+++ linux-5.12/drivers/i2c/busses/i2c-i801.c	2021-05-24 12:04:10.230998259 +0200
@@ -401,11 +401,9 @@ static int i801_check_post(struct i801_p
 		dev_err(&priv->pci_dev->dev, "Transaction timeout\n");
 		/* try to stop the current command */
 		dev_dbg(&priv->pci_dev->dev, "Terminating the current operation\n");
-		outb_p(inb_p(SMBHSTCNT(priv)) | SMBHSTCNT_KILL,
-		       SMBHSTCNT(priv));
+		outb_p(SMBHSTCNT_KILL, SMBHSTCNT(priv));
 		usleep_range(1000, 2000);
-		outb_p(inb_p(SMBHSTCNT(priv)) & (~SMBHSTCNT_KILL),
-		       SMBHSTCNT(priv));
+		outb_p(0, SMBHSTCNT(priv));
 
 		/* Check if it worked */
 		status = inb_p(SMBHSTSTS(priv));

-- 
Jean Delvare
SUSE L3 Support

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] i2c: i801: Don't generate an interrupt on bus reset
  2021-05-25 15:03 [PATCH] i2c: i801: Don't generate an interrupt on bus reset Jean Delvare
@ 2021-05-26  9:10 ` Jarkko Nikula
  2021-05-27 19:56 ` Wolfram Sang
  1 sibling, 0 replies; 3+ messages in thread
From: Jarkko Nikula @ 2021-05-26  9:10 UTC (permalink / raw)
  To: Jean Delvare, Linux I2C; +Cc: Andy Shevchenko, Heiner Kallweit

Hi

On 5/25/21 6:03 PM, Jean Delvare wrote:
> Now that the i2c-i801 driver supports interrupts, setting the KILL bit
> in a attempt to recover from a timed out transaction triggers an
> interrupt. Unfortunately, the interrupt handler (i801_isr) is not
> prepared for this situation and will try to process the interrupt as
> if it was signaling the end of a successful transaction. In the case
> of a block transaction, this can result in an out-of-range memory
> access.
> 
> This condition was reproduced several times by syzbot:
> https://syzkaller.appspot.com/bug?extid=ed71512d469895b5b34e
> https://syzkaller.appspot.com/bug?extid=8c8dedc0ba9e03f6c79e
> https://syzkaller.appspot.com/bug?extid=c8ff0b6d6c73d81b610e
> https://syzkaller.appspot.com/bug?extid=33f6c360821c399d69eb
> https://syzkaller.appspot.com/bug?extid=be15dc0b1933f04b043a
> https://syzkaller.appspot.com/bug?extid=b4d3fd1dfd53e90afd79
> 
> So disable interrupts while trying to reset the bus. Interrupts will
> be enabled again for the following transaction.
> 
> Fixes: 636752bcb517 ("i2c-i801: Enable IRQ for SMBus transactions")
> Reported-by: syzbot+b4d3fd1dfd53e90afd79@syzkaller.appspotmail.com
> Signed-off-by: Jean Delvare <jdelvare@suse.de>
> Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> Cc: Jarkko Nikula <jarkko.nikula@linux.intel.com>
> ---
>   drivers/i2c/busses/i2c-i801.c |    6 ++----
>   1 file changed, 2 insertions(+), 4 deletions(-)
> 
I hope I managed to test this properly. At least I was able to trigger 
KASAN quite easily and after your patch not anymore.

My test was to run "i2cdump -y 0 0x50 i" a few times with this hack:

--- a/drivers/i2c/busses/i2c-i801.c
+++ b/drivers/i2c/busses/i2c-i801.c
@@ -653,6 +653,7 @@ static irqreturn_t i801_isr(int irq, void *dev_id)
  	struct i801_priv *priv = dev_id;
  	u16 pcists;
  	u8 status;
+	static int count = 0;

  	/* Confirm this is our interrupt */
  	pci_read_config_word(priv->pci_dev, SMBPCISTS, &pcists);
@@ -676,8 +677,10 @@ static irqreturn_t i801_isr(int irq, void *dev_id)
  	status &= SMBHSTSTS_INTR | STATUS_ERROR_FLAGS;
  	if (status) {
  		outb_p(status, SMBHSTSTS(priv));
-		priv->status = status;
-		wake_up(&priv->waitq);
+		if ((count++ % 10)) {
+				priv->status = status;
+				wake_up(&priv->waitq);
+		}
  	}

  	return IRQ_HANDLED;

Tested-by: Jarkko Nikula <jarkko.nikula@linux.intel.com>

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] i2c: i801: Don't generate an interrupt on bus reset
  2021-05-25 15:03 [PATCH] i2c: i801: Don't generate an interrupt on bus reset Jean Delvare
  2021-05-26  9:10 ` Jarkko Nikula
@ 2021-05-27 19:56 ` Wolfram Sang
  1 sibling, 0 replies; 3+ messages in thread
From: Wolfram Sang @ 2021-05-27 19:56 UTC (permalink / raw)
  To: Jean Delvare; +Cc: Linux I2C, Andy Shevchenko, Jarkko Nikula, Heiner Kallweit

[-- Attachment #1: Type: text/plain, Size: 1421 bytes --]

On Tue, May 25, 2021 at 05:03:36PM +0200, Jean Delvare wrote:
> Now that the i2c-i801 driver supports interrupts, setting the KILL bit
> in a attempt to recover from a timed out transaction triggers an
> interrupt. Unfortunately, the interrupt handler (i801_isr) is not
> prepared for this situation and will try to process the interrupt as
> if it was signaling the end of a successful transaction. In the case
> of a block transaction, this can result in an out-of-range memory
> access.
> 
> This condition was reproduced several times by syzbot:
> https://syzkaller.appspot.com/bug?extid=ed71512d469895b5b34e
> https://syzkaller.appspot.com/bug?extid=8c8dedc0ba9e03f6c79e
> https://syzkaller.appspot.com/bug?extid=c8ff0b6d6c73d81b610e
> https://syzkaller.appspot.com/bug?extid=33f6c360821c399d69eb
> https://syzkaller.appspot.com/bug?extid=be15dc0b1933f04b043a
> https://syzkaller.appspot.com/bug?extid=b4d3fd1dfd53e90afd79
> 
> So disable interrupts while trying to reset the bus. Interrupts will
> be enabled again for the following transaction.
> 
> Fixes: 636752bcb517 ("i2c-i801: Enable IRQ for SMBus transactions")
> Reported-by: syzbot+b4d3fd1dfd53e90afd79@syzkaller.appspotmail.com
> Signed-off-by: Jean Delvare <jdelvare@suse.de>
> Acked-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> Cc: Jarkko Nikula <jarkko.nikula@linux.intel.com>

Applied to for-current, thanks!


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2021-05-27 19:57 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-05-25 15:03 [PATCH] i2c: i801: Don't generate an interrupt on bus reset Jean Delvare
2021-05-26  9:10 ` Jarkko Nikula
2021-05-27 19:56 ` Wolfram Sang

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).