On Thu, Sep 12, 2024 at 03:02:03PM +0300, Jani Nikula wrote:
> On Tue, 10 Sep 2024, Ville Syrjala <ville.syrjala@xxxxxxxxxxxxxxx> wrote:
> > From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> >
> > The SPI VBT codepath only knows how to read 4 bytes at a time.
> > So to read the 2 byte vbt_size it masks out the unwanted msbs.
> > Hide that little implementation detail inside a new intel_spi_read16()
> > helper. Alse rename the existing intel_spi_read() to intel_spi_read32()
> > to make it clear what it does.
> >
> > Signed-off-by: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx>
> > ---
> > drivers/gpu/drm/i915/display/intel_bios.c | 16 ++++++++++------
> > 1 file changed, 10 insertions(+), 6 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/i915/display/intel_bios.c b/drivers/gpu/drm/i915/display/intel_bios.c
> > index cc4a4cc2bf3e..cbbda94c3dab 100644
> > --- a/drivers/gpu/drm/i915/display/intel_bios.c
> > +++ b/drivers/gpu/drm/i915/display/intel_bios.c
> > @@ -3053,13 +3053,18 @@ static struct vbt_header *firmware_get_vbt(struct intel_display *display,
> > return vbt;
> > }
> >
> > -static u32 intel_spi_read(struct intel_uncore *uncore, u32 offset)
> > +static u32 intel_spi_read32(struct intel_uncore *uncore, u32 offset)
> > {
> > intel_uncore_write(uncore, PRIMARY_SPI_ADDRESS, offset);
> >
> > return intel_uncore_read(uncore, PRIMARY_SPI_TRIGGER);
> > }
> >
> > +static u16 intel_spi_read16(struct intel_uncore *uncore, u32 offset)
> > +{
> > + return intel_spi_read32(uncore, offset) & 0xffff;
> > +}
> > +
> > static struct vbt_header *spi_oprom_get_vbt(struct intel_display *display,
> > size_t *size)
> > {
> > @@ -3078,7 +3083,7 @@ static struct vbt_header *spi_oprom_get_vbt(struct intel_display *display,
> > oprom_offset &= OROM_OFFSET_MASK;
> >
> > for (count = 0; count < oprom_size; count += 4) {
> > - data = intel_spi_read(&i915->uncore, oprom_offset + count);
> > + data = intel_spi_read32(&i915->uncore, oprom_offset + count);
> > if (data == *((const u32 *)"$VBT")) {
> > found = oprom_offset + count;
> > break;
> > @@ -3094,9 +3099,8 @@ static struct vbt_header *spi_oprom_get_vbt(struct intel_display *display,
> > }
> >
> > /* Get VBT size and allocate space for the VBT */
> > - vbt_size = intel_spi_read(&i915->uncore,
> > - found + offsetof(struct vbt_header, vbt_size));
> > - vbt_size &= 0xffff;
> > + vbt_size = intel_spi_read16(&i915->uncore,
> > + found + offsetof(struct vbt_header, vbt_size));
>
> Pedantically if vbt_size was the last member of struct vbt_header this
> could read past the checked size, but it's not and meh. Also nothing to
> do with this change, apart from this hiding the detail. Still meh.
This code raises a lot of other lingering questions as well:
- do 8/16 bit accesses not work at all?
- what happens on an unaligned 32bit access?
>
> Reviewed-by: Jani Nikula <jani.nikula@xxxxxxxxx>
>
>
> >
> > if (vbt_size > oprom_size - count) {
> > drm_dbg(display->drm,
> > @@ -3109,7 +3113,7 @@ static struct vbt_header *spi_oprom_get_vbt(struct intel_display *display,
> > goto err_not_found;
> >
> > for (count = 0; count < vbt_size; count += 4)
> > - *(vbt + store++) = intel_spi_read(&i915->uncore, found + count);
> > + *(vbt + store++) = intel_spi_read32(&i915->uncore, found + count);
> >
> > if (!intel_bios_is_valid_vbt(display, vbt, vbt_size))
> > goto err_free_vbt;
>
> --
> Jani Nikula, Intel
--
Ville Syrjälä
Intel
