Re: [PATCH V2] ahci: Marvell 88SE9215 controllers prefer DMA for ATAPI

From: Huacai Chen
Date: Wed Mar 12 2025 - 09:34:52 EST

On Tue, Mar 11, 2025 at 6:45 PM Niklas Cassel <cassel@xxxxxxxxxx> wrote:
>
> Hello Huacai Chen,
>
> On Tue, Mar 11, 2025 at 11:02:17AM +0800, Huacai Chen wrote:
> > We use CD/DVD drives under Marvell 88SE9215 SATA controller on many
> > Loongson-based machines. We found its PIO doesn't work well, and on the
> > opposite its DMA seems work very well. We don't know the detail of the
> > 88SE9215 SATA controller, but we have tested different CD/DVD drives
> > and they all have problems under 88SE9215 (but they all work well under
> > an Intel SATA controller). So we can define a new AHCI board id named
> > board_ahci_atapi_dma, and for this id we set the ATA_FLAG_ATAPI_DMA and
> > ATA_QUIRK_ATAPI_MOD16_DMA flags on the SATA controller to prefer ATAPI
> > DMA.
>
> This patch does not apply.
> It conflicts with:
> https://web.git.kernel.org/pub/scm/linux/kernel/git/libata/linux.git/commit/?h=for-6.15&id=885251dc35767b1c992f6909532ca366c830814a
>
> Please add Daniel to CC when you respin, so that he might also be able
> to test.
OK.

>
>
> >
> > BTW, return -EOPNOTSUPP instead of 1 if ATAPI DMA is not supported in
> > atapi_check_dma().
>
> Please create a separate patch (e.g patch 1/2) for this with a proper
> commit log. (A proper commit log should always answer the question: Why?)
OK.

>
>
> >
> > Reported-by: Yuli Wang <wangyuli@xxxxxxxxxxxxx>
> > Tested-by: Jie Fan <fanjie@xxxxxxxxxxxxx>
> > Tested-by: Erpeng Xu <xuerpeng@xxxxxxxxxxxxx>
> > Tested-by: Yuli Wang <wangyuli@xxxxxxxxxxxxx>
> > Signed-off-by: Huacai Chen <chenhuacai@xxxxxxxxxxx>
> > ---
> > drivers/ata/ahci.c | 12 ++++++++++++
> > drivers/ata/libata-core.c | 6 +++++-
> > include/linux/libata.h | 1 +
> > 3 files changed, 18 insertions(+), 1 deletion(-)
> >
> > diff --git a/drivers/ata/ahci.c b/drivers/ata/ahci.c
> > index f813dbdc2346..a64db28549d8 100644
> > --- a/drivers/ata/ahci.c
> > +++ b/drivers/ata/ahci.c
> > @@ -49,6 +49,7 @@ enum board_ids {
> > /* board IDs by feature in alphabetical order */
> > board_ahci,
> > board_ahci_43bit_dma,
> > + board_ahci_atapi_dma,
> > board_ahci_ign_iferr,
> > board_ahci_no_debounce_delay,
> > board_ahci_no_msi,
> > @@ -137,6 +138,12 @@ static const struct ata_port_info ahci_port_info[] = {
> > .udma_mask = ATA_UDMA6,
> > .port_ops = &ahci_ops,
> > },
> > + [board_ahci_atapi_dma] = {
> > + .flags = AHCI_FLAG_COMMON,
> > + .pio_mask = ATA_PIO4,
> > + .udma_mask = ATA_UDMA6,
> > + .port_ops = &ahci_ops,
> > + },
> > [board_ahci_ign_iferr] = {
> > AHCI_HFLAGS (AHCI_HFLAG_IGN_IRQ_IF_ERR),
> > .flags = AHCI_FLAG_COMMON,
> > @@ -591,6 +598,8 @@ static const struct pci_device_id ahci_pci_tbl[] = {
> > .driver_data = board_ahci_yes_fbs },
> > { PCI_DEVICE(PCI_VENDOR_ID_MARVELL_EXT, 0x9230),
> > .driver_data = board_ahci_yes_fbs },
> > + { PCI_DEVICE(PCI_VENDOR_ID_MARVELL_EXT, 0x9215),
> > + .driver_data = board_ahci_atapi_dma },
> > { PCI_DEVICE(PCI_VENDOR_ID_MARVELL_EXT, 0x9235),
> > .driver_data = board_ahci_no_debounce_delay },
> > { PCI_DEVICE(PCI_VENDOR_ID_TTI, 0x0642), /* highpoint rocketraid 642L */
> > @@ -1917,6 +1926,9 @@ static int ahci_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
> > /* save initial config */
> > ahci_pci_save_initial_config(pdev, hpriv);
> >
> > + if (board_id == board_ahci_atapi_dma)
> > + pi.flags |= ATA_FLAG_ATAPI_DMA;
> > +
>
> No need for these three lines, just rename your board_ahci_atapi_dma
> board to board_ahci_atapi_dma_quirk_yes_fbs
OK, but I will use a different name such as
board_ahci_yes_fbs_atapi_dma since yes_fbs is the main feature for
marvell controllers.

>
> and in the initialization of ahci_atapi_dma_quirk_yes_fbs, set:
> AHCI_HFLAGS (AHCI_HFLAG_ATAPI_DMA_QUIRK |
> AHCI_HFLAG_YES_FBS),
> .flags = AHCI_FLAG_COMMON,
>
>
> And rename ATA_FLAG_ATAPI_DMA to AHCI_HFLAG_ATAPI_DMA_QUIRK and put it in
> AHCI_HFLAGS instead.
OK.

>
>
>
> > /* prepare host */
> > if (hpriv->cap & HOST_CAP_NCQ) {
> > pi.flags |= ATA_FLAG_NCQ;
> > diff --git a/drivers/ata/libata-core.c b/drivers/ata/libata-core.c
> > index c085dd81ebe7..87a3dbf3ac93 100644
> > --- a/drivers/ata/libata-core.c
> > +++ b/drivers/ata/libata-core.c
> > @@ -3029,6 +3029,10 @@ int ata_dev_configure(struct ata_device *dev)
> > dev->max_sectors = ATA_MAX_SECTORS;
> > }
> >
> > + if ((dev->class == ATA_DEV_ATAPI) &&
> > + (ap->flags & ATA_FLAG_ATAPI_DMA))
> > + dev->quirks |= ATA_QUIRK_ATAPI_MOD16_DMA;
> > +
> > if ((dev->class == ATA_DEV_ATAPI) &&
> > (atapi_command_packet_set(id) == TYPE_TAPE)) {
> > dev->max_sectors = ATA_MAX_SECTORS_TAPE;
> > @@ -4544,7 +4548,7 @@ int atapi_check_dma(struct ata_queued_cmd *qc)
> > */
> > if (!(qc->dev->quirks & ATA_QUIRK_ATAPI_MOD16_DMA) &&
> > unlikely(qc->nbytes & 15))
> > - return 1;
> > + return -EOPNOTSUPP;
> >
> > if (ap->ops->check_atapi_dma)
> > return ap->ops->check_atapi_dma(qc);
> > diff --git a/include/linux/libata.h b/include/linux/libata.h
> > index c1c57f814b98..67d374279a65 100644
> > --- a/include/linux/libata.h
> > +++ b/include/linux/libata.h
> > @@ -194,6 +194,7 @@ enum {
> > /* (doesn't imply presence) */
> > ATA_FLAG_SATA = (1 << 1),
> > ATA_FLAG_NO_LPM = (1 << 2), /* host not happy with LPM */
> > + ATA_FLAG_ATAPI_DMA = (1 << 4), /* ATAPI use DMA */
>
> s/ATAPI use DMA/force ATAPI to use DMA/
OK.

Huacai
>
>
> > ATA_FLAG_NO_LOG_PAGE = (1 << 5), /* do not issue log page read */
> > ATA_FLAG_NO_ATAPI = (1 << 6), /* No ATAPI support */
> > ATA_FLAG_PIO_DMA = (1 << 7), /* PIO cmds via DMA */
> > --
> > 2.47.1
> >
>
> Kind regards,
> Niklas