| Message ID | 20210420084226.2996654-1-narmstrong@baylibre.com |
|---|---|
| Headers | show |
| Series | reset: fix reset_get_by_index_nodev index handling | expand |
Hi Tom, Simon, On 20/04/2021 10:42, Neil Armstrong wrote: > A regression weas detected on Amlogic G12A/G12B SoCs, where HDMI output was disable > even when Linux was booting. > > Bisect reports 139e4a1cbe ("drivers: reset: Add a managed API to get reset controllers from the DT") > as the offending commit. > > But the error is in ea9dc35aab ("reset: Get the RESET by index without device") where a spurius "> 0" > was added to the index handling. > > But the dm_test_reset_base() test did not catch it. > > The first commit extends the test to catch the regression, and the second patch fixes the regression. So you have any comment on the patch order here ? Should I push the fixed test after the fix ? Neil > > Neil Armstrong (2): > test: reset: Extend base reset test to catch error > reset: fix reset_get_by_index_nodev index handling > > arch/sandbox/dts/test.dts | 4 ++-- > drivers/reset/reset-uclass.c | 2 +- > test/dm/reset.c | 39 +++++++++++++++++++++++++++++++----- > 3 files changed, 37 insertions(+), 8 deletions(-) >
On Tue, Apr 27, 2021 at 09:37:11AM +0200, Neil Armstrong wrote: > Hi Tom, Simon, > > On 20/04/2021 10:42, Neil Armstrong wrote: > > A regression weas detected on Amlogic G12A/G12B SoCs, where HDMI output was disable > > even when Linux was booting. > > > > Bisect reports 139e4a1cbe ("drivers: reset: Add a managed API to get reset controllers from the DT") > > as the offending commit. > > > > But the error is in ea9dc35aab ("reset: Get the RESET by index without device") where a spurius "> 0" > > was added to the index handling. > > > > But the dm_test_reset_base() test did not catch it. > > > > The first commit extends the test to catch the regression, and the second patch fixes the regression. > > So you have any comment on the patch order here ? Should I push the fixed test after the fix ? I'll grab these shortly, thanks for the reminder.
A regression weas detected on Amlogic G12A/G12B SoCs, where HDMI output was disable even when Linux was booting. Bisect reports 139e4a1cbe ("drivers: reset: Add a managed API to get reset controllers from the DT") as the offending commit. But the error is in ea9dc35aab ("reset: Get the RESET by index without device") where a spurius "> 0" was added to the index handling. But the dm_test_reset_base() test did not catch it. The first commit extends the test to catch the regression, and the second patch fixes the regression. Neil Armstrong (2): test: reset: Extend base reset test to catch error reset: fix reset_get_by_index_nodev index handling arch/sandbox/dts/test.dts | 4 ++-- drivers/reset/reset-uclass.c | 2 +- test/dm/reset.c | 39 +++++++++++++++++++++++++++++++----- 3 files changed, 37 insertions(+), 8 deletions(-)