[PATCH v2 0/4] ahci: a bunch of small fixes.

[Gerd Hoffmann]

Hi,

Here is a bunch of small fixes for the ahci driver which accumulated
over the last week.

[ v2: use pci_config_maskw, one patch updated ]

cheers,
  Gerd

Gerd Hoffmann (4):
  ahci: set dma feature flag
  ahci: enable io/mem/dma
  ahci: fix off-by-one in port count
  ahci: set controller id

src/ahci.c |    9 ++++++++-
 1 files changed, 8 insertions(+), 1 deletions(-)

[PATCH v2 1/4] ahci: set dma feature flag

[Gerd Hoffmann]

Signed-off-by: Gerd Hoffmann 
---
 src/ahci.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/src/ahci.c b/src/ahci.c
index 6c3127a..ee50e8f 100644
--- a/src/ahci.c
+++ b/src/ahci.c
@@ -50,6 +50,7 @@ static void sata_prep_readwrite(struct sata_cmd_fis *fis,
         command = (iswrite ? ATA_CMD_WRITE_DMA
                    : ATA_CMD_READ_DMA);
     }
+    SET_FLATPTR(fis->feature,      1); /* dma */
     SET_FLATPTR(fis->command,      command);
     SET_FLATPTR(fis->sector_count, op->count);
     SET_FLATPTR(fis->lba_low,      lba);
@@ -62,6 +63,7 @@ static void sata_prep_atapi(struct sata_cmd_fis *fis, u16 blocksize)
 {
     memset_fl(fis, 0, sizeof(*fis));
     SET_FLATPTR(fis->command,  ATA_CMD_PACKET);
+    SET_FLATPTR(fis->feature,  1); /* dma */
     SET_FLATPTR(fis->lba_mid,  blocksize);
     SET_FLATPTR(fis->lba_high, blocksize >> 8);
 }

[PATCH v2 2/4] ahci: enable io/mem/dma

[Gerd Hoffmann]

Make sure IO, MMIO and DMA are enabled in pci config space
before using the device.

[ v2: use pci_config_maskw ]

Signed-off-by: Gerd Hoffmann 
---
 src/ahci.c |    4 ++++
 1 files changed, 4 insertions(+), 0 deletions(-)

diff --git a/src/ahci.c b/src/ahci.c
index ee50e8f..0b23f43 100644
--- a/src/ahci.c
+++ b/src/ahci.c
@@ -436,6 +436,7 @@ ahci_init_controller(int bdf)
 {
     struct ahci_ctrl_s *ctrl = malloc_fseg(sizeof(*ctrl));
     u32 val;
+    u16 cmd;

if (!ctrl) {
         warn_noalloc();
@@ -447,6 +448,9 @@ ahci_init_controller(int bdf)
     dprintf(1, "AHCI controller at %02x.%x, iobase %x, irq %d\n",
             bdf >> 3, bdf & 7, ctrl->iobase, ctrl->irq);

+    pci_config_maskw(bdf, PCI_COMMAND, 0,
+                     PCI_COMMAND_IO | PCI_COMMAND_MEMORY | PCI_COMMAND_MASTER);
+
     val = ahci_ctrl_readl(ctrl, HOST_CTL);
     ahci_ctrl_writel(ctrl, HOST_CTL, val | HOST_CTL_AHCI_EN);

[PATCH v2 2/4] ahci: enable io/mem/dma

[Kevin O'Connor]

On Thu, Dec 09, 2010 at 08:39:46AM +0100, Gerd Hoffmann wrote:
> Make sure IO, MMIO and DMA are enabled in pci config space
> before using the device.
>
> [ v2: use pci_config_maskw ]
> Signed-off-by: Gerd Hoffmann
>
> src/ahci.c |    4 ++++
>  1 files changed, 4 insertions(+), 0 deletions(-)
>
> diff --git a/src/ahci.c b/src/ahci.c
> index ee50e8f..0b23f43 100644
> --- a/src/ahci.c
> +++ b/src/ahci.c
> @@ -436,6 +436,7 @@ ahci_init_controller(int bdf)
>  {
>      struct ahci_ctrl_s *ctrl = malloc_fseg(sizeof(*ctrl));
>      u32 val;
> u16 cmd;

You have an unused variable here.

The patch series looks fine to me.  If there's no further comments
I'll commit this weekend.

-Kevin

[PATCH v2 2/4] ahci: enable io/mem/dma

[Gerd Hoffmann]

Hi,
> > u16 cmd;
>
> You have an unused variable here.

Ah, right.  Howcome gcc didn't warn me on this, usually it does ...
Is that warning turned off for seabios?
> The patch series looks fine to me.  If there's no further comments
> I'll commit this weekend.

Want me respin?  Do do you just zap the variable when committing?

cheers,
    Gerd

[PATCH v2 2/4] ahci: enable io/mem/dma

[Kevin O'Connor]

On Fri, Dec 10, 2010 at 10:33:28AM +0100, Gerd Hoffmann wrote:
> Hi,
> > > u16 cmd;
> >
> > You have an unused variable here.
>
> Ah, right.  Howcome gcc didn't warn me on this, usually it does ...
> Is that warning turned off for seabios?

It warns for me.

Compiling whole program out/ccode.16.s
src/ahci.c: In function ‘ahci_init_controller’:
src/ahci.c:437:9: warning: unused variable ‘cmd’

Maybe something with different gcc versions?

$ gcc --version
gcc (GCC) 4.5.1 20100924 (Red Hat 4.5.1-4)
> > The patch series looks fine to me.  If there's no further comments
> > I'll commit this weekend.
>
> Want me respin?  Do do you just zap the variable when committing?

I can fix it if it's easier.

-Kevin

[PATCH v2 3/4] ahci: fix off-by-one in port count

[Gerd Hoffmann]

Signed-off-by: Gerd Hoffmann 
---
 src/ahci.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/src/ahci.c b/src/ahci.c
index 0b23f43..7abff75 100644
--- a/src/ahci.c
+++ b/src/ahci.c
@@ -417,7 +417,7 @@ ahci_detect(void *data)
     int rc;

max = ctrl->caps & 0x1f;
-    for (pnr = 0; pnr < max; pnr++) {
+    for (pnr = 0; pnr <= max; pnr++) {
         if (!(ctrl->ports & (1 << pnr)))
             continue;
         dprintf(2, "AHCI/%d: probing\n", pnr);

[PATCH v2 4/4] ahci: set controller id

[Gerd Hoffmann]

Fill the controller id in the drive struct with the port number so we
get a sane boot menu ordering with multiple hard disks attached.

Signed-off-by: Gerd Hoffmann 
---
 src/ahci.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/src/ahci.c b/src/ahci.c
index 7abff75..5373e72 100644
--- a/src/ahci.c
+++ b/src/ahci.c
@@ -345,6 +345,7 @@ ahci_port_init(struct ahci_ctrl_s *ctrl, u32 pnr)
     }

port->drive.type = DTYPE_AHCI;
+    port->drive.cntl_id = pnr;
     port->drive.removable = (buffer[0] & 0x80) ? 1 : 0;
     port->drive.desc = malloc_tmp(MAXDESCSIZE);
     if (!port->drive.desc) {

[Kevin O'Connor]

On Thu, Dec 09, 2010 at 08:39:44AM +0100, Gerd Hoffmann wrote:
> Hi,
>
> Here is a bunch of small fixes for the ahci driver which accumulated
> over the last week.
>
> [ v2: use pci_config_maskw, one patch updated ]

I committed this series.

-Kevin