Re: SATA disk on cntroller in AHCI mode not found: Hang at `AHCI status

grub-devel

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: SATA disk on cntroller in AHCI mode not found: Hang at `AHCI status

From:	Paul Menzel
Subject:	Re: SATA disk on cntroller in AHCI mode not found: Hang at `AHCI status <1 0 150>`
Date:	Fri, 19 Apr 2013 16:40:16 +0200

Dear GRUB folks,


Am Donnerstag, den 18.04.2013, 13:11 +0200 schrieb Paul Menzel:

> here is an update on the issue I only reported to #grub on
> <irc.freenode.net>.
> 
> 
> Am Freitag, den 05.04.2013, 12:54 +0200 schrieb Paul Menzel:
> 
> > as reported to the coreboot mailing list already [1][2], using GRUB
> > (4778) as a payload [3] on the AMD SB800 based ASRock E350M1 [4], the
> > SATA disk attached to the SATA controller is not found when the
> > controller is in AHCI mode. Using IDE mode works.
> 
> […]
> 
> > I’ll gladly test patches and provide more information.
> 
> Vladmir gave me the following patch to test [5].

[…]

> The change to `ahci.c` did make a change and it progressed further.
> 
>         grub> insmod pata
>         disk/pata.c:295: sectors=0x5a
>         disk/pata.c:295: sectors=0x5a
>         disk/pata.c:295: sectors=0xff
>         error: no device connected.
>         disk/pata.c:295: sectors=0xff
>         error: no device connected.
>         grub> insmod ahci
>         disk/ahci.c:308: Requesting AHCI ownership
>         disk/ahci.c:311: Waiting for BIOS to give up ownership
>         disk/ahci.c:322: AHCI ownership obtained
>         disk/ahci.c:328: AHCI is in compat mode. Switching
>         disk/ahci.c:374: 6 AHCI ports
>         disk/ahci.c:384: status 0:123
>         disk/ahci.c:186: found device ahci0 (port 0)
>         disk/ahci.c:229: det = 3
>         disk/ahci.c:240: link up
>         disk/ahci.c:246: After PORT_CMD_ST
>         disk/ahci.c:255: before return 0
>         disk/ahci.c:384: status 1:0
>         disk/ahci.c:384: status 2:0
>         disk/ahci.c:384: status 3:0
>         disk/ahci.c:384: status 4:0
>         disk/ahci.c:384: status 5:0
>         grub> ls
>         disk/ahci.c:710: opening AHCI dev `ahci0'
>         disk/ahci.c:535: AHCI tfd = 150
>         disk/ahci.c:561: grub_ahci_read (size=512, cmdsize = 0)
>         disk/ahci.c:574: AHCI tfd = 150
>         disk/ahci.c:604: cfis: 27 80 ec 00 00 00 00 e0
>         disk/ahci.c:609: cfis: 00 00 00 00 00 00 00 00
>         disk/ahci.c:622: PRDT = c7eed800, 0, 800001ff (128)
>         disk/ahci.c:627: AHCI command schedulded
>         disk/ahci.c:629: AHCI tfd = 150
>         disk/ahci.c:633: AHCI tfd = 150
>         disk/ahci.c:635: AHCI sig = 101
>         disk/ahci.c:637: AHCI tfd = 150
>         disk/ahci.c:646: AHCI status <1 0 150>
>         disk/ahci.c:658: AHCI command completed <1 0 150 0 0, 0 0>
>         disk/ahci.c:668: last PIO FIS 00000000 00000000 00000000 00000000 
> 00000000
>         00000000 00000000 00000000
>         disk/ahci.c:678: last REG FIS 00000000 00000000 00000000 00000000 
> 00000000
>         00000000 00000000 00000000
>         disk/pata.c:295: sectors=0x5a
>         disk/pata.c:295: sectors=0x5a
>         disk/pata.c:295: sectors=0xff
>         disk/pata.c:295: sectors=0xff
>         disk/ahci.c:710: opening AHCI dev `ahci0'
>         disk/ahci.c:535: AHCI tfd = 150
>         disk/ahci.c:561: grub_ahci_read (size=512, cmdsize = 0)
>         disk/ahci.c:574: AHCI tfd = 150
>         disk/ahci.c:604: cfis: 27 80 ec 00 00 00 00 e0
>         disk/ahci.c:609: cfis: 00 00 00 00 00 00 00 00
>         disk/ahci.c:622: PRDT = c7eed800, 0, 800001ff (128)
>         disk/ahci.c:627: AHCI command schedulded
>         disk/ahci.c:629: AHCI tfd = 150
>         disk/ahci.c:633: AHCI tfd = 150
>         disk/ahci.c:635: AHCI sig = 101
>         disk/ahci.c:637: AHCI tfd = 150
>         disk/ahci.c:646: AHCI status <1 0 150>
>         disk/ahci.c:658: AHCI command completed <1 0 150 0 0, 0 0>
>         disk/ahci.c:668: last PIO FIS 00000000 00000000 00000000 00000000 
> 00000000
>         00000000 00000000 00000000
>         disk/ahci.c:678: last REG FIS 00000000 00000000 00000000 00000000 
> 00000000
>         00000000 00000000 00000000
>         disk/pata.c:295: sectors=0x5a
>         disk/pata.c:295: sectors=0x5a
>         disk/pata.c:295: sectors=0xff
>         disk/pata.c:295: sectors=0xff
>         (memdisk)
> 
> So it gets further but still does not find the disk.
> 
> Vladimir also said, he can reproduce this issue, when putting his AMD
> board’s SATA controller in IDE mode in the BIOS and then trying to
> change it in GRUB to AHCI mode.

Vladimir gave me another patch to try changing the reset code [6].

        === modified file 'grub-core/disk/ahci.c'
        --- grub-core/disk/ahci.c       2010-12-25 11:11:54 +0000
        +++ grub-core/disk/ahci.c       2010-12-25 15:30:53 +0000
        @@ -131,7 +131,8 @@
         
         static grub_err_t 
         grub_ahci_readwrite_real (struct grub_ahci_device *dev,
        -                         struct grub_disk_ata_pass_through_parms 
*parms);
        +                         struct grub_disk_ata_pass_through_parms 
*parms,
        +                         int reset);
         
         
         enum
        @@ -374,9 +375,30 @@
                                              9 /* LBA48 mid */,
                                              10 /* LBA48 high */ }; 
         
        +static grub_err_t
        +grub_ahci_reset_port (struct grub_ahci_device *dev, int force)
        +{
        +  dev->hba->ports[dev->port].sata_error = 
dev->hba->ports[dev->port].sata_error;
        +
        +  if (force || (dev->hba->ports[dev->port].command_issue & 1)
        +      || (dev->hba->ports[dev->port].task_file_data & 0x80))
        +    {
        +      struct grub_disk_ata_pass_through_parms parms2;
        +      dev->hba->ports[dev->port].command &= ~GRUB_AHCI_HBA_PORT_CMD_ST;
        +      while ((dev->hba->ports[dev->port].command & 
GRUB_AHCI_HBA_PORT_CMD_CR));
        +      dev->hba->ports[dev->port].command |= GRUB_AHCI_HBA_PORT_CMD_ST;
        +      while (!(dev->hba->ports[dev->port].command & 
GRUB_AHCI_HBA_PORT_CMD_CR));
        +      dev->hba->ports[dev->port].sata_error = 
dev->hba->ports[dev->port].sata_error;
        +      grub_memset (&parms2, 0, sizeof (parms2));
        +      parms2.taskfile.cmd = 8;
        +      return grub_ahci_readwrite_real (dev, &parms2, 1);
        +    }
        +  return GRUB_ERR_NONE;
        +}
        +
         static grub_err_t 
         grub_ahci_readwrite_real (struct grub_ahci_device *dev,
        -                         struct grub_disk_ata_pass_through_parms 
*parms)
        +                         struct grub_disk_ata_pass_through_parms 
*parms, int reset)
         {
           struct grub_pci_dma_chunk *bufc;
           grub_uint64_t endtime;
        @@ -386,15 +408,8 @@
           grub_dprintf ("ahci", "AHCI tfd = %x\n",
                        dev->hba->ports[dev->port].task_file_data);
         
        -  if ((dev->hba->ports[dev->port].task_file_data & 0x80))
        -    {
        -      dev->hba->ports[dev->port].command &= ~GRUB_AHCI_HBA_PORT_CMD_ST;
        -      while ((dev->hba->ports[dev->port].command & 
GRUB_AHCI_HBA_PORT_CMD_CR));
        -      dev->hba->ports[dev->port].command |= GRUB_AHCI_HBA_PORT_CMD_ST;
        -      while (!(dev->hba->ports[dev->port].command & 
GRUB_AHCI_HBA_PORT_CMD_CR));
        -    }
        -
        -  dev->hba->ports[dev->port].sata_error = 
dev->hba->ports[dev->port].sata_error;
        +  if (!reset)
        +    grub_ahci_reset_port (dev, 0);
         
           grub_dprintf("ahci", "grub_ahci_read (size=%llu, cmdsize = %llu)\n",
                       (unsigned long long) parms->size,
        @@ -475,7 +490,7 @@
           grub_dprintf ("ahci", "AHCI tfd = %x\n",
                        dev->hba->ports[dev->port].task_file_data);
         
        -  endtime = grub_get_time_ms () + 1000;
        +  endtime = grub_get_time_ms () + 10000;
           while ((dev->hba->ports[dev->port].command_issue & 1))
             if (grub_get_time_ms () > endtime)
               {
        @@ -483,7 +498,8 @@
                              dev->hba->ports[dev->port].command_issue,
                              dev->hba->ports[dev->port].intstatus,
                              dev->hba->ports[dev->port].task_file_data);
        -       err = grub_error (GRUB_ERR_IO, "AHCI transfer timeouted");
        +       grub_ahci_reset_port (dev, 1);
        +       err = grub_error (GRUB_ERR_IO, "AHCI transfer timed out");
                break;
               }
         
        @@ -527,7 +543,7 @@
         grub_ahci_readwrite (grub_ata_t disk,
                             struct grub_disk_ata_pass_through_parms *parms)
         {
        -  return grub_ahci_readwrite_real (disk->data, parms);
        +  return grub_ahci_readwrite_real (disk->data, parms, 0);
         }
         
         static grub_err_t

It does not apply cleanly to my version and misses probably something as
`grub_ahci_readwrite` was also changed but not `ata.h`. I am attaching
my adapted version of `ahci.c` I tested. Here is the result.
Unfortunately it hangs now.

        grub> insmod ahci
        disk/ahci.c:308: Requesting AHCI ownership
        disk/ahci.c:311: Waiting for BIOS to give up ownership
        disk/ahci.c:322: AHCI ownership obtained
        disk/ahci.c:328: AHCI is in compat mode. Switching
        disk/ahci.c:374: 6 AHCI ports
        disk/ahci.c:384: status 0:123
        disk/ahci.c:186: found device ahci0 (port 0)
        disk/ahci.c:229: det = 3
        disk/ahci.c:240: link up
        disk/ahci.c:246: After PORT_CMD_ST
        disk/ahci.c:255: before return 0
        disk/ahci.c:384: status 1:0
        disk/ahci.c:384: status 2:0
        disk/ahci.c:384: status 3:0
        disk/ahci.c:384: status 4:0
        disk/ahci.c:384: status 5:0
        grub> ls
        disk/ahci.c:713: opening AHCI dev `ahci0'
        disk/ahci.c:555: AHCI tfd = 150
        disk/ahci.c:562: grub_ahci_read (size=512, cmdsize = 0)
        disk/ahci.c:575: AHCI tfd = 150
        disk/ahci.c:605: cfis: 27 80 ec 00 00 00 00 e0
        disk/ahci.c:610: cfis: 00 00 00 00 00 00 00 00
        disk/ahci.c:623: PRDT = c7eedc00, 0, 800001ff (128)
        disk/ahci.c:628: AHCI command schedulded
        disk/ahci.c:630: AHCI tfd = 150
        disk/ahci.c:634: AHCI tfd = 150
        disk/ahci.c:636: AHCI sig = 101
        disk/ahci.c:638: AHCI tfd = 150
        disk/ahci.c:647: AHCI status <1 0 150>

I have to reset the machine to break the hang.

Hopefully I did not mess up applying the patch. If yes, it would be nice
to test some branch with the latest AHCI test code.


Thanks,

Paul


> > [1] http://www.coreboot.org/pipermail/coreboot/2013-March/075583.html
> > [2] http://www.coreboot.org/pipermail/coreboot/2013-April/075592.html
> > [3] http://www.coreboot.org/GRUB2
> > [4] http://www.coreboot.org/ASRock_E350M1
> [5] http://paste.debian.net/249135/
[6] http://paste.debian.net/249577/

ahci.c
Description: Text Data

signature.asc
Description: This is a digitally signed message part

[Prev in Thread]

Current Thread

[Next in Thread]

SATA disk on cntroller in AHCI mode not found: `disk/ahci.c:225: couldn't start CR`, Paul Menzel, 2013/04/05
- Re: SATA disk on cntroller in AHCI mode not found: `disk/ahci.c:225: couldn't start CR`, Paul Menzel, 2013/04/18
  - Re: SATA disk on cntroller in AHCI mode not found: Hang at `AHCI status <1 0 150>`, Paul Menzel <=

Prev by Date: Re: [3/4] [PATCH] start angle (option of circular_progress) is measured in unusual quantities
Next by Date: Re: [PATCH] Fix ofpathname for sas disks
Previous by thread: Re: SATA disk on cntroller in AHCI mode not found: `disk/ahci.c:225: couldn't start CR`
Next by thread: [PATCH] make autogen.sh work on tardist again
Index(es):
- Date
- Thread