2020-03-11 11:31:23

by Bean Huo

Subject: [PATCH 0/1] Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"

From: Bean Huo <[email protected]>

Hi, Martin

Based on Bart's feedack, the less risky way is to revert commit:

34656dda81ac "scsi: ufs: Let the SCSI core allocate per-command UFS data"

Bean Huo (1):
Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"

drivers/scsi/ufs/ufshcd.c | 198 ++++++++++++++------------------------
drivers/scsi/ufs/ufshcd.h | 5 +
2 files changed, 75 insertions(+), 128 deletions(-)

--
2.17.1

2020-03-11 12:26:40

by Avri Altman

Subject: RE: [PATCH 0/1] Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"

Hi,

>
> From: Bean Huo <[email protected]>
>
> Hi, Martin
>
> Based on Bart's feedack, the less risky way is to revert commit:
>
> 34656dda81ac "scsi: ufs: Let the SCSI core allocate per-command UFS data"
>
> Bean Huo (1):
> Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"
Maybe it's safer to revert the entire series?

Thanks,
Avri


>
> drivers/scsi/ufs/ufshcd.c | 198 ++++++++++++++------------------------
> drivers/scsi/ufs/ufshcd.h | 5 +
> 2 files changed, 75 insertions(+), 128 deletions(-)
>
> --
> 2.17.1

2020-03-11 12:53:59

by Bean Huo (beanhuo)

Subject: RE: [EXT] RE: [PATCH 0/1] Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"

Hi, Avri
Do you want to revert all or two (scsi: ufs: Let the SCSI core allocate per-command UFS data, scsi: core: Introduce {init,exit}_cmd_priv() )?
Because the patches "scsi: ufs: Simplify two tests" and "scsi: ufs: Introduce ufshcd_init_lrb()" are ok to me.
No problem keeping them. Just this one "scsi: core: Introduce {init,exit}_cmd_priv()" is not necessary, since no drivers it now.

//Bean


>
> Hi,
>
> >
> > From: Bean Huo <[email protected]>
> >
> > Hi, Martin
> >
> > Based on Bart's feedack, the less risky way is to revert commit:
> >
> > 34656dda81ac "scsi: ufs: Let the SCSI core allocate per-command UFS data"
> >
> > Bean Huo (1):
> > Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"
> Maybe it's safer to revert the entire series?
>
> Thanks,
> Avri
>
>
> >
> > drivers/scsi/ufs/ufshcd.c | 198 ++++++++++++++------------------------
> > drivers/scsi/ufs/ufshcd.h | 5 +
> > 2 files changed, 75 insertions(+), 128 deletions(-)
> >
> > --
> > 2.17.1

2020-03-11 14:06:03

by Bart Van Assche

Subject: Re: [EXT] RE: [PATCH 0/1] Revert "scsi: ufs: Let the SCSI core allocate per-command UFS data"

On 3/11/20 5:53 AM, Bean Huo (beanhuo) wrote:
> Hi, Avri
> Do you want to revert all or two (scsi: ufs: Let the SCSI core allocate per-command UFS data, scsi: core: Introduce {init,exit}_cmd_priv() )?
> Because the patches "scsi: ufs: Simplify two tests" and "scsi: ufs: Introduce ufshcd_init_lrb()" are ok to me.
> No problem keeping them. Just this one "scsi: core: Introduce {init,exit}_cmd_priv()" is not necessary, since no drivers it now.

I agree with Bean. The other patches from the same series are small and
easy to review. The SCSI core patch has been tested with other SCSI LLDs.

Bart.