On Tue, May 30, 2023 at 11:06:55AM +0200, Paolo Abeni wrote:
> On Mon, 2023-05-29 at 21:58 -0700, Jakub Kicinski wrote:
> > On Fri, 26 May 2023 14:45:54 +0300 Dan Carpenter wrote:
> > > The "val" variable is used to store error codes from phy_read() so
> > > it needs to be signed for the error handling to work as expected.
> > >
> > > Fixes: 014068dcb5b1 ("net: phy: genphy_loopback: add link speed configuration")
> > > Signed-off-by: Dan Carpenter <dan.carpenter@xxxxxxxxxx>
> >
> > Is it going to be obvious to PHY-savvy folks that the val passed to
> > phy_read_poll_timeout() must be an int? Is it a very common pattern?
> > My outsider intuition is that since regs are 16b, u16 is reasonable,
> > and more people may make the same mistake. Therefore we should try to
> > fix phy_read_poll_timeout() instead to use a local variable like it
> > does for __ret.
> >
> > Weaker version would be to add a compile time check to ensure val
> > is signed (assert(typeof(val)~0ULL < 0) or such?).
>
> FTR, a BUILD_BUG_ON() the above check spots issues in several places
> (e.g. r8169_main.c, drivers/net/phy/phy_device.c, ...)
>
I don't see an issue in r8169_main.c and in drivers/net/phy/phy_device.c
then I only find the bug from this patch.
regards,
dan carpenter
diff --git a/include/linux/phy.h b/include/linux/phy.h
index 6478838405a08..f05fc25b77583 100644
--- a/include/linux/phy.h
+++ b/include/linux/phy.h
@@ -1173,6 +1173,7 @@ static inline int phy_read(struct phy_device *phydev, u32 regnum)
({ \
int __ret = read_poll_timeout(phy_read, val, val < 0 || (cond), \
sleep_us, timeout_us, sleep_before_read, phydev, regnum); \
+ BUILD_BUG_ON((typeof(val))~0ULL > 0); \
if (val < 0) \
__ret = val; \
if (__ret) \
