| Message ID | 1459865146-25308-1-git-send-email-john@metanate.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Am Dienstag, 5. April 2016, 15:05:46 schrieb John Keeping: > Since commit cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi > slaves") rk_iommu_is_stall_active() always returns false because the > bitwise AND operates on the boolean flag promoted to an integer and a > value that is either zero or BIT(2). > > Explicitly convert the right-hand value to a boolean so that both sides > are guaranteed to be either zero or one. > > rk_iommu_is_paging_enabled() does not suffer from the same problem since > RK_MMU_STATUS_PAGING_ENABLED is BIT(0), but let's apply the same change > for consistency and to make it clear that it's correct without needing > to lookup the value. > > Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi > slaves") Signed-off-by: John Keeping <john@metanate.com> Nice catch, thanks. Reviewed-by: Heiko Stuebner <heiko@sntech.de> John, out of curiosity, how did you find that problem? Excess stall error messages or something else? Thanks Heiko > --- > drivers/iommu/rockchip-iommu.c | 8 ++++---- > 1 file changed, 4 insertions(+), 4 deletions(-) > > diff --git a/drivers/iommu/rockchip-iommu.c > b/drivers/iommu/rockchip-iommu.c index a6f593a0a29e..5710a06c3049 100644 > --- a/drivers/iommu/rockchip-iommu.c > +++ b/drivers/iommu/rockchip-iommu.c > @@ -315,8 +315,8 @@ static bool rk_iommu_is_stall_active(struct rk_iommu > *iommu) int i; > > for (i = 0; i < iommu->num_mmu; i++) > - active &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > - RK_MMU_STATUS_STALL_ACTIVE; > + active &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > + RK_MMU_STATUS_STALL_ACTIVE); > > return active; > } > @@ -327,8 +327,8 @@ static bool rk_iommu_is_paging_enabled(struct rk_iommu > *iommu) int i; > > for (i = 0; i < iommu->num_mmu; i++) > - enable &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > - RK_MMU_STATUS_PAGING_ENABLED; > + enable &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > + RK_MMU_STATUS_PAGING_ENABLED); > > return enable; > }
On Tue, 05 Apr 2016 10:21:27 -0700, Heiko Stuebner wrote: > Am Dienstag, 5. April 2016, 15:05:46 schrieb John Keeping: > > Since commit cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi > > slaves") rk_iommu_is_stall_active() always returns false because the > > bitwise AND operates on the boolean flag promoted to an integer and a > > value that is either zero or BIT(2). > > > > Explicitly convert the right-hand value to a boolean so that both sides > > are guaranteed to be either zero or one. > > > > rk_iommu_is_paging_enabled() does not suffer from the same problem since > > RK_MMU_STATUS_PAGING_ENABLED is BIT(0), but let's apply the same change > > for consistency and to make it clear that it's correct without needing > > to lookup the value. > > > > Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi > > slaves") Signed-off-by: John Keeping <john@metanate.com> > > Nice catch, thanks. > > Reviewed-by: Heiko Stuebner <heiko@sntech.de> > > John, out of curiosity, how did you find that problem? Excess stall error > messages or something else? I was testing the drm iommu patches I posted today [1] and noticed an "Enable stall request timed out" error when unbinding the display-subsystem, but the RK_MMU_STATUS value printed in the error message indicated that stall active was set. [1] http://article.gmane.org/gmane.comp.video.dri.devel/150721 > > --- > > drivers/iommu/rockchip-iommu.c | 8 ++++---- > > 1 file changed, 4 insertions(+), 4 deletions(-) > > > > diff --git a/drivers/iommu/rockchip-iommu.c > > b/drivers/iommu/rockchip-iommu.c index a6f593a0a29e..5710a06c3049 100644 > > --- a/drivers/iommu/rockchip-iommu.c > > +++ b/drivers/iommu/rockchip-iommu.c > > @@ -315,8 +315,8 @@ static bool rk_iommu_is_stall_active(struct rk_iommu > > *iommu) int i; > > > > for (i = 0; i < iommu->num_mmu; i++) > > - active &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > - RK_MMU_STATUS_STALL_ACTIVE; > > + active &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > + RK_MMU_STATUS_STALL_ACTIVE); > > > > return active; > > } > > @@ -327,8 +327,8 @@ static bool rk_iommu_is_paging_enabled(struct rk_iommu > > *iommu) int i; > > > > for (i = 0; i < iommu->num_mmu; i++) > > - enable &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > - RK_MMU_STATUS_PAGING_ENABLED; > > + enable &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > + RK_MMU_STATUS_PAGING_ENABLED); > > > > return enable; > > } >
Am Dienstag, 5. April 2016, 18:28:58 schrieb John Keeping: > On Tue, 05 Apr 2016 10:21:27 -0700, Heiko Stuebner wrote: > > Am Dienstag, 5. April 2016, 15:05:46 schrieb John Keeping: > > > Since commit cd6438c5f844 ("iommu/rockchip: Reconstruct to support > > > multi > > > slaves") rk_iommu_is_stall_active() always returns false because the > > > bitwise AND operates on the boolean flag promoted to an integer and a > > > value that is either zero or BIT(2). > > > > > > Explicitly convert the right-hand value to a boolean so that both > > > sides > > > are guaranteed to be either zero or one. > > > > > > rk_iommu_is_paging_enabled() does not suffer from the same problem > > > since > > > RK_MMU_STATUS_PAGING_ENABLED is BIT(0), but let's apply the same > > > change > > > for consistency and to make it clear that it's correct without needing > > > to lookup the value. > > > > > > Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi > > > slaves") Signed-off-by: John Keeping <john@metanate.com> > > > > Nice catch, thanks. > > > > Reviewed-by: Heiko Stuebner <heiko@sntech.de> > > > > John, out of curiosity, how did you find that problem? Excess stall > > error > > messages or something else? > > I was testing the drm iommu patches I posted today [1] and noticed an > "Enable stall request timed out" error when unbinding the > display-subsystem, but the RK_MMU_STATUS value printed in the error > message indicated that stall active was set. > > [1] http://article.gmane.org/gmane.comp.video.dri.devel/150721 So it seems both you and Tomeu were working to resolve the same issue [2]. And John's patch seems to actually fix the underlying problem. [2] https://lkml.org/lkml/2016/3/22/493 > > > --- > > > > > > drivers/iommu/rockchip-iommu.c | 8 ++++---- > > > 1 file changed, 4 insertions(+), 4 deletions(-) > > > > > > diff --git a/drivers/iommu/rockchip-iommu.c > > > b/drivers/iommu/rockchip-iommu.c index a6f593a0a29e..5710a06c3049 > > > 100644 > > > --- a/drivers/iommu/rockchip-iommu.c > > > +++ b/drivers/iommu/rockchip-iommu.c > > > @@ -315,8 +315,8 @@ static bool rk_iommu_is_stall_active(struct > > > rk_iommu > > > *iommu) int i; > > > > > > for (i = 0; i < iommu->num_mmu; i++) > > > > > > - active &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > > - RK_MMU_STATUS_STALL_ACTIVE; > > > + active &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > > + RK_MMU_STATUS_STALL_ACTIVE); > > > > > > return active; > > > > > > } > > > > > > @@ -327,8 +327,8 @@ static bool rk_iommu_is_paging_enabled(struct > > > rk_iommu *iommu) int i; > > > > > > for (i = 0; i < iommu->num_mmu; i++) > > > > > > - enable &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > > - RK_MMU_STATUS_PAGING_ENABLED; > > > + enable &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & > > > + RK_MMU_STATUS_PAGING_ENABLED); > > > > > > return enable; > > > > > > }
On 5 April 2016 at 21:10, Heiko Stuebner <heiko@sntech.de> wrote: > Am Dienstag, 5. April 2016, 18:28:58 schrieb John Keeping: >> On Tue, 05 Apr 2016 10:21:27 -0700, Heiko Stuebner wrote: >> > Am Dienstag, 5. April 2016, 15:05:46 schrieb John Keeping: >> > > Since commit cd6438c5f844 ("iommu/rockchip: Reconstruct to support >> > > multi >> > > slaves") rk_iommu_is_stall_active() always returns false because the >> > > bitwise AND operates on the boolean flag promoted to an integer and a >> > > value that is either zero or BIT(2). >> > > >> > > Explicitly convert the right-hand value to a boolean so that both >> > > sides >> > > are guaranteed to be either zero or one. >> > > >> > > rk_iommu_is_paging_enabled() does not suffer from the same problem >> > > since >> > > RK_MMU_STATUS_PAGING_ENABLED is BIT(0), but let's apply the same >> > > change >> > > for consistency and to make it clear that it's correct without needing >> > > to lookup the value. >> > > >> > > Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi >> > > slaves") Signed-off-by: John Keeping <john@metanate.com> >> > >> > Nice catch, thanks. >> > >> > Reviewed-by: Heiko Stuebner <heiko@sntech.de> >> > >> > John, out of curiosity, how did you find that problem? Excess stall >> > error >> > messages or something else? >> >> I was testing the drm iommu patches I posted today [1] and noticed an >> "Enable stall request timed out" error when unbinding the >> display-subsystem, but the RK_MMU_STATUS value printed in the error >> message indicated that stall active was set. >> >> [1] http://article.gmane.org/gmane.comp.video.dri.devel/150721 > > So it seems both you and Tomeu were working to resolve the same issue [2]. > > And John's patch seems to actually fix the underlying problem. Indeed, I don't get spurious messages any more with this patch. Tested-by: Tomeu Vizoso <tomeu.vizoso@collabora.com> Thanks, Tomeu > > [2] https://lkml.org/lkml/2016/3/22/493 > > >> > > --- >> > > >> > > drivers/iommu/rockchip-iommu.c | 8 ++++---- >> > > 1 file changed, 4 insertions(+), 4 deletions(-) >> > > >> > > diff --git a/drivers/iommu/rockchip-iommu.c >> > > b/drivers/iommu/rockchip-iommu.c index a6f593a0a29e..5710a06c3049 >> > > 100644 >> > > --- a/drivers/iommu/rockchip-iommu.c >> > > +++ b/drivers/iommu/rockchip-iommu.c >> > > @@ -315,8 +315,8 @@ static bool rk_iommu_is_stall_active(struct >> > > rk_iommu >> > > *iommu) int i; >> > > >> > > for (i = 0; i < iommu->num_mmu; i++) >> > > >> > > - active &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & >> > > - RK_MMU_STATUS_STALL_ACTIVE; >> > > + active &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & >> > > + RK_MMU_STATUS_STALL_ACTIVE); >> > > >> > > return active; >> > > >> > > } >> > > >> > > @@ -327,8 +327,8 @@ static bool rk_iommu_is_paging_enabled(struct >> > > rk_iommu *iommu) int i; >> > > >> > > for (i = 0; i < iommu->num_mmu; i++) >> > > >> > > - enable &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & >> > > - RK_MMU_STATUS_PAGING_ENABLED; >> > > + enable &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & >> > > + RK_MMU_STATUS_PAGING_ENABLED); >> > > >> > > return enable; >> > > >> > > } >
On Tue, Apr 05, 2016 at 03:05:46PM +0100, John Keeping wrote: > drivers/iommu/rockchip-iommu.c | 8 ++++---- > 1 file changed, 4 insertions(+), 4 deletions(-) Applied to iommu/fixes. Thanks.
diff --git a/drivers/iommu/rockchip-iommu.c b/drivers/iommu/rockchip-iommu.c index a6f593a0a29e..5710a06c3049 100644 --- a/drivers/iommu/rockchip-iommu.c +++ b/drivers/iommu/rockchip-iommu.c @@ -315,8 +315,8 @@ static bool rk_iommu_is_stall_active(struct rk_iommu *iommu) int i; for (i = 0; i < iommu->num_mmu; i++) - active &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & - RK_MMU_STATUS_STALL_ACTIVE; + active &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & + RK_MMU_STATUS_STALL_ACTIVE); return active; } @@ -327,8 +327,8 @@ static bool rk_iommu_is_paging_enabled(struct rk_iommu *iommu) int i; for (i = 0; i < iommu->num_mmu; i++) - enable &= rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & - RK_MMU_STATUS_PAGING_ENABLED; + enable &= !!(rk_iommu_read(iommu->bases[i], RK_MMU_STATUS) & + RK_MMU_STATUS_PAGING_ENABLED); return enable; }
Since commit cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi slaves") rk_iommu_is_stall_active() always returns false because the bitwise AND operates on the boolean flag promoted to an integer and a value that is either zero or BIT(2). Explicitly convert the right-hand value to a boolean so that both sides are guaranteed to be either zero or one. rk_iommu_is_paging_enabled() does not suffer from the same problem since RK_MMU_STATUS_PAGING_ENABLED is BIT(0), but let's apply the same change for consistency and to make it clear that it's correct without needing to lookup the value. Fixes: cd6438c5f844 ("iommu/rockchip: Reconstruct to support multi slaves") Signed-off-by: John Keeping <john@metanate.com> --- drivers/iommu/rockchip-iommu.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-)