Re: [PATCH v2 1/6] zorro_esp: Limit DMA transfers to 65535 bytes

[Finn Thain]

On Sun, 14 Oct 2018, Geert Uytterhoeven wrote:

> 
> > Fixes: 3109e5ae0311
> 
> Fixes: 3109e5ae0311 ("scsi: zorro_esp: New driver for Amiga Zorro
> NCR53C9x boards")
> 

OK.

> $ git help fixes
> 'fixes' is aliased to 'show --format='Fixes: %h ("%s")' -s'
> 
> BTW, if you use vim, add
> 
>     noremap ;gs "zyiw:exe "new \| setlocal buftype=nofile
> bufhidden=hide noswapfile syntax=git \| 0r ! git show ".@z."" \|
> :0<CR><CR>
> 
> to .vimrc, and type ";gs" when the cursor is positioned on a commit ID.
> 

Thanks.

> > Signed-off-by: Finn Thain <fthain@xxxxxxxxxxxxxxxxxxx>
> > Cc: Michael Schmitz <schmitzmic@xxxxxxxxx>
> > Tested-by: Michael Schmitz <schmitzmic@xxxxxxxxx>
> 
> > --- a/drivers/scsi/zorro_esp.c
> > +++ b/drivers/scsi/zorro_esp.c
> > @@ -245,7 +245,7 @@ static int fastlane_esp_irq_pending(struct esp *esp)
> >  static u32 zorro_esp_dma_length_limit(struct esp *esp, u32 dma_addr,
> >                                         u32 dma_len)
> >  {
> > -       return dma_len > 0xFFFFFF ? 0xFFFFFF : dma_len;
> > +       return dma_len > 0xFFFF ? 0xFFFF : dma_len;
> >  }
> >
> >  static void zorro_esp_reset_dma(struct esp *esp)
> > @@ -484,7 +484,6 @@ static void zorro_esp_send_blz1230_dma_cmd(struct esp *esp, u32 addr,
> >         scsi_esp_cmd(esp, ESP_CMD_DMA);
> >         zorro_esp_write8(esp, (esp_count >> 0) & 0xff, ESP_TCLOW);
> >         zorro_esp_write8(esp, (esp_count >> 8) & 0xff, ESP_TCMED);
> > -       zorro_esp_write8(esp, (esp_count >> 16) & 0xff, ESP_TCHI);
> 
> Since all these sub-drivers seem to support 24-bit transfers, perhaps this is
> something to be fixed in the esp_scsi core?
> 

I don't think there is anything to fix in the esp_scsi core. The fact that 
no-one saw the error indicates that large transfers don't occur in 
practice.

If you said there is an opportunity for an enhancement, I could agree, 
assuming that you also found a way to produce large IO requests. But I 
doubt that such an enhancement would make a measurable improvement.

Even if the benefit was measurable, the fix under review would still be 
needed for stable kernels.

BTW, Michael and I already discussed all this (it probably should have 
happened on the linux-m68k list).

Please see,

$ grep -lr ESP_CONFIG2_FENAB drivers/scsi/
drivers/scsi/am53c974.c
drivers/scsi/esp_scsi.c
drivers/scsi/esp_scsi.h

The authors of am53c974.c obviously decided it wasn't wise to make this 
feature mandatory (which suggests that perhaps it shouldn't go into common 
code).

The comments in esp_scsi.c say that ESP_CONFIG2_FENAB has undesirable 
consequences. There is information in the datasheets on this point but I 
didn't follow it up because I don't see a performance issue.

--