[PATCH v2] drm/ast: astdp: fix loop timeout check
Dan Carpenter
dan.carpenter at linaro.org
Mon Aug 12 06:54:41 UTC 2024
On Mon, Aug 12, 2024 at 08:48:16AM +0200, Thomas Zimmermann wrote:
> Hi
>
> Am 12.08.24 um 08:42 schrieb Dan Carpenter:
> > This code has an issue because it loops until "i" is set to UINT_MAX but
> > the test for failure assumes that "i" is set to zero. The result is that
> > it will only print an error message if we succeed on the very last try.
> > Reformat the loop to count forwards instead of backwards.
> >
> > Fixes: 2281475168d2 ("drm/ast: astdp: Perform link training during atomic_enable")
> > Signed-off-by: Dan Carpenter <dan.carpenter at linaro.org>
> > ---
> > v2: In version one, I introduced a bug where it would msleep(100) after failure
> > and that is a pointless thing to do. Also change the loop to a for loop.
> > ---
> > drivers/gpu/drm/ast/ast_dp.c | 12 +++++-------
> > 1 file changed, 5 insertions(+), 7 deletions(-)
> >
> > diff --git a/drivers/gpu/drm/ast/ast_dp.c b/drivers/gpu/drm/ast/ast_dp.c
> > index 5d07678b502c..9bc21dd6a54d 100644
> > --- a/drivers/gpu/drm/ast/ast_dp.c
> > +++ b/drivers/gpu/drm/ast/ast_dp.c
> > @@ -146,18 +146,16 @@ void ast_dp_power_on_off(struct drm_device *dev, bool on)
> > void ast_dp_link_training(struct ast_device *ast)
> > {
> > struct drm_device *dev = &ast->base;
> > - unsigned int i = 10;
> > + int i;
> > - while (i--) {
> > + for (i = 0; i < 10; i++) {
> > u8 vgacrdc = ast_get_index_reg(ast, AST_IO_VGACRI, 0xdc);
> > if (vgacrdc & AST_IO_VGACRDC_LINK_SUCCESS)
> > - break;
> > - if (i)
> > - msleep(100);
> > + return;
> > + msleep(100);
>
> But we don't want to wait during the final iteration of this loop. If you
> want to use the for loop, it should be something like
>
> for (i= 0; i < 10; ++i) {
>
> if (i)
> msleep(100)
>
> // now test vgacrdc
> }
>
> Best regards
> Thomas
I feel like if we really hit this failure path then we won't care about the
tenth msleep(). I can resend if you want, but I'd prefer to just leave it.
regards,
dan carpenter
More information about the dri-devel
mailing list