[V1,0/2] Fix hibern8 enter failure during host reset

Message ID 20230725192710.26698-1-quic_nitirawa@quicinc.com
Headers
Series Fix hibern8 enter failure during host reset |

Message

Nitin Rawat July 25, 2023, 7:27 p.m. UTC
  Commit <52a518019c> ("Fix missing clk change notification on host reset")
added clock scaling notification to ufshcd_host_reset_and_restore.
This invokes hibern8 enter and exit on qualcomm platform which fails
because controller is in reset state.

Fix this by checking the Host controller state before sending
hibern8 command.

__ufshcd_wl_resume()
ufshcd_reset_and_restore()
ufshcd_host_reset_and_restore()
ufshcd_scale_clks()
ufshcd_vops_clk_scale_notify()
ufs_qcom_clk_scale_notify()
ufshcd_uic_hibern8_enter()

Fixes: 52a518019ca1 ("scsi: ufs: core: Fix missing clk change notification on host reset")

Nitin Rawat (2):
  scsi: ufs: core: Export ufshcd_is_hba_active
  scsi: ufs: ufs-qcom: check host controller state

 drivers/ufs/core/ufshcd.c   | 3 ++-
 drivers/ufs/host/ufs-qcom.c | 4 ++++
 include/ufs/ufshcd.h        | 1 +
 3 files changed, 7 insertions(+), 1 deletion(-)

--
2.17.1
  

Comments

Bjorn Andersson July 26, 2023, 5:41 a.m. UTC | #1
On Wed, Jul 26, 2023 at 12:57:08AM +0530, Nitin Rawat wrote:
> Commit <52a518019c> ("Fix missing clk change notification on host reset")
> added clock scaling notification to ufshcd_host_reset_and_restore.
> This invokes hibern8 enter and exit on qualcomm platform which fails
> because controller is in reset state.
> 
> Fix this by checking the Host controller state before sending
> hibern8 command.
> 
> __ufshcd_wl_resume()
> ufshcd_reset_and_restore()
> ufshcd_host_reset_and_restore()
> ufshcd_scale_clks()
> ufshcd_vops_clk_scale_notify()
> ufs_qcom_clk_scale_notify()
> ufshcd_uic_hibern8_enter()
> 
> Fixes: 52a518019ca1 ("scsi: ufs: core: Fix missing clk change notification on host reset")

Sorry, I missed this before I looked at the individual patches.

The cover-letter is not included in the git history, so this information
will be lost. This explains the reason for patch 2, so please put it
there.

Thanks,
Bjorn