Hi, On Wed, Mar 29, 2023 at 04:54:25PM -0400, Adrien Thierry wrote: > This reverts commit 7dafc3e007918384c8693ff8d70381b5c1e9c247. > > This patch introduced a regression [1] where hba->pwr_info is used > before being initialized, which could create issues in > ufshcd_scale_gear(). Revert it until a better solution is found. > > [1] https://lore.kernel.org/all/CAGaU9a_PMZhqv+YJ0r3w-hJMsR922oxW6Kg59vw+oen-NZ6Otw@xxxxxxxxxxxxxx > > Signed-off-by: Adrien Thierry <athierry@xxxxxxxxxx> > --- > drivers/ufs/core/ufshcd.c | 47 +++++++++++++-------------------------- > include/ufs/ufshcd.h | 1 - > 2 files changed, 16 insertions(+), 32 deletions(-) > I've been working on a fixed version, and realized the original issue [1] has been fixed by [2]. Thanks to the softdep, whenever the ufs core and the simple ondemand governor are built as modules, the governor module will be loaded before the ufs core module. Thus, the ufs core doesn't end up calling request_module because the governor is already loaded. So, it looks like there's now no need to submit a fixed version. [1] https://lore.kernel.org/all/20230217194423.42553-1-athierry@xxxxxxxxxx/ [2] https://lore.kernel.org/all/20230220140740.14379-1-athierry@xxxxxxxxxx/ Best, Adrien
