On Mon, Oct 14, 2024 at 08:16:47PM +0200, Borislav Petkov wrote:
> On Fri, Oct 11, 2024 at 11:31:31AM -0400, Frank Li wrote:
> > From: Priyanka Singh <priyanka.singh@xxxxxxx>
> >
> > Fix undefined behavior caused by left-shifting a negative value in the
> > expression:
> >
> > cap_high ^ (1 << (bad_data_bit - 32))
> >
> > The variable `bad_data_bit` ranges from 0 to 63. When `bad_data_bit` is
> > less than 32, `bad_data_bit - 32` becomes negative, and left-shifting by a
> > negative value in C is undefined behavior.
> >
> > Fix this by checking the range of `bad_data_bit` before performing the
> > shift.
> >
> > Fixes: ea2eb9a8b620 ("EDAC, fsl-ddr: Separate FSL DDR driver from MPC85xx")
>
> Is this an urgent fix which needs to go to stable or someone just caught it
> from code review?
I don't think it is urgent. In most system the return value is 0. I am not
sure who caught it because patch already exist at downstream tree for a
whole.
>
> Does it trigger in real life, IOW?
The problem is triggered. But the output result is correct at our hardware.
The result may change depend on compiler and cpu version.
Frank
>
> > Signed-off-by: Priyanka Singh <priyanka.singh@xxxxxxx>
> > Reviewed-by: Sherry Sun <sherry.sun@xxxxxxx>
> > Signed-off-by: Li Yang <leoyang.li@xxxxxxx>
> > Signed-off-by: Frank Li <Frank.Li@xxxxxxx>
> > ---
> > drivers/edac/fsl_ddr_edac.c | 17 ++++++++++++-----
> > 1 file changed, 12 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/edac/fsl_ddr_edac.c b/drivers/edac/fsl_ddr_edac.c
> > index 7a9fb1202f1a0..ccc13c2adfd6f 100644
> > --- a/drivers/edac/fsl_ddr_edac.c
> > +++ b/drivers/edac/fsl_ddr_edac.c
> > @@ -338,11 +338,18 @@ static void fsl_mc_check(struct mem_ctl_info *mci)
> > fsl_mc_printk(mci, KERN_ERR,
> > "Faulty ECC bit: %d\n", bad_ecc_bit);
> >
> > - fsl_mc_printk(mci, KERN_ERR,
> > - "Expected Data / ECC:\t%#8.8x_%08x / %#2.2x\n",
> > - cap_high ^ (1 << (bad_data_bit - 32)),
> > - cap_low ^ (1 << bad_data_bit),
> > - syndrome ^ (1 << bad_ecc_bit));
> > + if ((bad_data_bit > 0 && bad_data_bit < 32) && bad_ecc_bit > 0) {
> > + fsl_mc_printk(mci, KERN_ERR,
> > + "Expected Data / ECC:\t%#8.8x_%08x / %#2.2x\n",
> > + cap_high, cap_low ^ (1 << bad_data_bit),
> > + syndrome ^ (1 << bad_ecc_bit));
> > + }
> > + if (bad_data_bit >= 32 && bad_ecc_bit > 0) {
> > + fsl_mc_printk(mci, KERN_ERR,
> > + "Expected Data / ECC:\t%#8.8x_%08x / %#2.2x\n",
> > + cap_high ^ (1 << (bad_data_bit - 32)),
> > + cap_low, syndrome ^ (1 << bad_ecc_bit));
> > + }
>
> This is getting unnecessarily clumsy than it should be. Please do the
> following:
>
> if (bad_data_bit != 1 && bad_ecc_bit != -1) {
>
> // prep the values you need to print
>
> // do an exactly one fsl_mc_printk() with the prepared values.
>
> }
>
> Not have 4 fsl_mc_printks with a bunch of silly if-checks in front.
>
> Thx.
>
> --
> Regards/Gruss,
> Boris.
>
> https://people.kernel.org/tglx/notes-about-netiquette
