Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port 2-4

[Carl-Daniel Hailfinger]

Hi Zheng,

On 23.12.2008 03:26, Bao, Zheng wrote:
> Reviewed-by: zheng bao
>
> It is tested well on dbm690t and pistachio.

Thank you for the fast review!

Our commit process is a bit difficult because it needs an Acked-by:
statement. (The subversion repository will reject a commit without
Acked-by.)

Everybody who understands the code is allowed to ack a patch. You
understand that piece of code very well, so could you please also add an
Acked-by statement? Thanks!

Regards,
Carl-Daniel
> -----Original Message-----
> From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
> Sent: Tuesday, December 23, 2008 10:01 AM
> To: Coreboot
> Cc: Bao, Zheng
> Subject: [PATCH] Fix SB600 SATA and add support for port 2-4
>
> The SB600 RPR documentation does not mention what to do if SATA_BAR0+6
> is no longer 0xA0 or 0xB0. It simply assumes that will never happen.
> My 500 GB Seagate Barracuda ST3500820AS triggers that corner case on the
> first init after poweron.
> The current code hangs forever with my drive. Fix this by rerunning the
> init sequence after SATA_BAR0+6 is no longer 0xA0 or 0xB0.
>
> Add support for SATA port 2-4 (Primary Slave, Secondary Master,
> Secondary Slave).
>
> Activate and improve debug messages for SPEW log level.
>
> Fix some comments.
>
> New log messages look like this:
> PCI: 00:12.0 init
> sata_bar0=3020
> sata_bar1=3060
> sata_bar2=3030
> sata_bar3=3070
> sata_bar4=3000
> sata_bar5=fc309000
> SATA port 0 status = 23
> 0x6=a0, 0x7=80
> drive detection not yet completed, waiting...
> 0x6=a0, 0x7=80
> drive detection not yet completed, waiting...
> [... 281 repetitions ...]
> 0x6=0, 0x7=50
> drive no longer selected after 2830 ms, retrying init
> drive detection done after 10 ms
> Primary Master device is ready
> SATA port 1 status = 23
> drive detection done after 10 ms
> Primary Slave device is ready
> SATA port 2 status = 0
> No Secondary Master SATA drive on Slot2
> SATA port 3 status = 0
> No Secondary Slave SATA drive on Slot3
>
> Full log is attached.
>
> With this patch (and my other non-SATA fixups), my Asus M2A-VM boots
> into Linux without problems.
>
> Signed-off-by: Carl-Daniel Hailfinger

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

Testing is a rapid process. I need some time to truly review it to
understand your code. Acked-by will be added later.

Zheng

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Tuesday, December 23, 2008 10:37 AM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [PATCH] Fix SB600 SATA and add support for port 2-4

Hi Zheng,

On 23.12.2008 03:26, Bao, Zheng wrote:
> Reviewed-by: zheng bao
>
> It is tested well on dbm690t and pistachio.

Thank you for the fast review!

Our commit process is a bit difficult because it needs an Acked-by:
statement. (The subversion repository will reject a commit without
Acked-by.)

Everybody who understands the code is allowed to ack a patch. You
understand that piece of code very well, so could you please also add an
Acked-by statement? Thanks!

Regards,
Carl-Daniel
> -----Original Message-----
> From: Carl-Daniel Hailfinger

[mailto:[e-mail]]
> Sent: Tuesday, December 23, 2008 10:01 AM
> To: Coreboot
> Cc: Bao, Zheng
> Subject: [PATCH] Fix SB600 SATA and add support for port 2-4
>
> The SB600 RPR documentation does not mention what to do if SATA_BAR0+6
> is no longer 0xA0 or 0xB0. It simply assumes that will never happen.
> My 500 GB Seagate Barracuda ST3500820AS triggers that corner case on

the
> first init after poweron.
> The current code hangs forever with my drive. Fix this by rerunning

the
> init sequence after SATA_BAR0+6 is no longer 0xA0 or 0xB0.
>
> Add support for SATA port 2-4 (Primary Slave, Secondary Master,
> Secondary Slave).
>
> Activate and improve debug messages for SPEW log level.
>
> Fix some comments.
>
> New log messages look like this:
> PCI: 00:12.0 init
> sata_bar0=3020
> sata_bar1=3060
> sata_bar2=3030
> sata_bar3=3070
> sata_bar4=3000
> sata_bar5=fc309000
> SATA port 0 status = 23
> 0x6=a0, 0x7=80
> drive detection not yet completed, waiting...
> 0x6=a0, 0x7=80
> drive detection not yet completed, waiting...
> [... 281 repetitions ...]
> 0x6=0, 0x7=50
> drive no longer selected after 2830 ms, retrying init
> drive detection done after 10 ms
> Primary Master device is ready
> SATA port 1 status = 23
> drive detection done after 10 ms
> Primary Slave device is ready
> SATA port 2 status = 0
> No Secondary Master SATA drive on Slot2
> SATA port 3 status = 0
> No Secondary Slave SATA drive on Slot3
>
> Full log is attached.
>
> With this patch (and my other non-SATA fixups), my Asus M2A-VM boots
> into Linux without problems.
>
> Signed-off-by: Carl-Daniel Hailfinger

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

In sata_drive_detect(), the loop seems to be loop forever if detection
is not completed. Does the "i" have a timeout value?

Some indents are actually several spaces instead of tab. Please note
this.

Zheng

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Tuesday, December 23, 2008 10:37 AM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [PATCH] Fix SB600 SATA and add support for port 2-4

Hi Zheng,

On 23.12.2008 03:26, Bao, Zheng wrote:
> Reviewed-by: zheng bao
>
> It is tested well on dbm690t and pistachio.

Thank you for the fast review!

Our commit process is a bit difficult because it needs an Acked-by:
statement. (The subversion repository will reject a commit without
Acked-by.)

Everybody who understands the code is allowed to ack a patch. You
understand that piece of code very well, so could you please also add an
Acked-by statement? Thanks!

Regards,
Carl-Daniel
> -----Original Message-----
> From: Carl-Daniel Hailfinger

[mailto:[e-mail]]
> Sent: Tuesday, December 23, 2008 10:01 AM
> To: Coreboot
> Cc: Bao, Zheng
> Subject: [PATCH] Fix SB600 SATA and add support for port 2-4
>
> The SB600 RPR documentation does not mention what to do if SATA_BAR0+6
> is no longer 0xA0 or 0xB0. It simply assumes that will never happen.
> My 500 GB Seagate Barracuda ST3500820AS triggers that corner case on

the
> first init after poweron.
> The current code hangs forever with my drive. Fix this by rerunning

the
> init sequence after SATA_BAR0+6 is no longer 0xA0 or 0xB0.
>
> Add support for SATA port 2-4 (Primary Slave, Secondary Master,
> Secondary Slave).
>
> Activate and improve debug messages for SPEW log level.
>
> Fix some comments.
>
> New log messages look like this:
> PCI: 00:12.0 init
> sata_bar0=3020
> sata_bar1=3060
> sata_bar2=3030
> sata_bar3=3070
> sata_bar4=3000
> sata_bar5=fc309000
> SATA port 0 status = 23
> 0x6=a0, 0x7=80
> drive detection not yet completed, waiting...
> 0x6=a0, 0x7=80
> drive detection not yet completed, waiting...
> [... 281 repetitions ...]
> 0x6=0, 0x7=50
> drive no longer selected after 2830 ms, retrying init
> drive detection done after 10 ms
> Primary Master device is ready
> SATA port 1 status = 23
> drive detection done after 10 ms
> Primary Slave device is ready
> SATA port 2 status = 0
> No Secondary Master SATA drive on Slot2
> SATA port 3 status = 0
> No Secondary Slave SATA drive on Slot3
>
> Full log is attached.
>
> With this patch (and my other non-SATA fixups), my Asus M2A-VM boots
> into Linux without problems.
>
> Signed-off-by: Carl-Daniel Hailfinger

[PATCH] Fix SB600 SATA and add support for port 2-4

[Carl-Daniel Hailfinger]

On 23.12.2008 04:51, Bao, Zheng wrote:
> In sata_drive_detect(), the loop seems to be loop forever if detection
> is not completed. Does the "i" have a timeout value?

No, there is no timeout. "i" was there to tell me how long it takes, but
I can use it to check for a timeout. I will fix this.
> Some indents are actually several spaces instead of tab. Please note
> this.

Thanks for telling me. I will clean up and resend. Some indents have
tabs and spaces to have exact indentation at the beginning of the
parentheses.

Regards,
Carl-Daniel
> Zheng
>
> -----Original Message-----
> From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
> Sent: Tuesday, December 23, 2008 10:37 AM
> To: Bao, Zheng
> Cc: Coreboot
> Subject: Re: [PATCH] Fix SB600 SATA and add support for port 2-4
>
> Hi Zheng,
>
> On 23.12.2008 03:26, Bao, Zheng wrote:
> > Reviewed-by: zheng bao
> >
> > It is tested well on dbm690t and pistachio.
>
> Thank you for the fast review!
>
> Our commit process is a bit difficult because it needs an Acked-by:
> statement. (The subversion repository will reject a commit without
> Acked-by.)
>
> Everybody who understands the code is allowed to ack a patch. You
> understand that piece of code very well, so could you please also add an
> Acked-by statement? Thanks!
>
> Regards,
> Carl-Daniel
> > -----Original Message-----
> > From: Carl-Daniel Hailfinger
>
> [mailto:[e-mail]]
> > Sent: Tuesday, December 23, 2008 10:01 AM
> > To: Coreboot
> > Cc: Bao, Zheng
> > Subject: [PATCH] Fix SB600 SATA and add support for port 2-4
> >
> > The SB600 RPR documentation does not mention what to do if SATA_BAR0+6
> > is no longer 0xA0 or 0xB0. It simply assumes that will never happen.
> > My 500 GB Seagate Barracuda ST3500820AS triggers that corner case on
>
> the
> > first init after poweron.
> > The current code hangs forever with my drive. Fix this by rerunning
>
> the
> > init sequence after SATA_BAR0+6 is no longer 0xA0 or 0xB0.
> >
> > Add support for SATA port 2-4 (Primary Slave, Secondary Master,
> > Secondary Slave).
> >
> > Activate and improve debug messages for SPEW log level.
> >
> > Fix some comments.
> >
> > New log messages look like this:
> > PCI: 00:12.0 init
> > sata_bar0=3020
> > sata_bar1=3060
> > sata_bar2=3030
> > sata_bar3=3070
> > sata_bar4=3000
> > sata_bar5=fc309000
> > SATA port 0 status = 23
> > 0x6=a0, 0x7=80
> > drive detection not yet completed, waiting...
> > 0x6=a0, 0x7=80
> > drive detection not yet completed, waiting...
> > [... 281 repetitions ...]
> > 0x6=0, 0x7=50
> > drive no longer selected after 2830 ms, retrying init
> > drive detection done after 10 ms
> > Primary Master device is ready
> > SATA port 1 status = 23
> > drive detection done after 10 ms
> > Primary Slave device is ready
> > SATA port 2 status = 0
> > No Secondary Master SATA drive on Slot2
> > SATA port 3 status = 0
> > No Secondary Slave SATA drive on Slot3
> >
> > Full log is attached.
> >
> > With this patch (and my other non-SATA fixups), my Asus M2A-VM boots
> > into Linux without problems.
> >
> > Signed-off-by: Carl-Daniel Hailfinger

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

Sorry, I tested it again and tried it on 4 ports. It only works on 1st
and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
driver no longer selected after 10ms, retrying init
driver no longer selected after 10ms, retrying init
driver no longer selected after 10ms, retrying init
driver no longer selected after 10ms, retrying init
driver no longer selected after 10ms, retrying init

My SATA drive is 250 GB Seagate Barracuda ST3250620NS.

Zheng

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Tuesday, December 23, 2008 12:01 PM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port
2-4

On 23.12.2008 04:51, Bao, Zheng wrote:
> In sata_drive_detect(), the loop seems to be loop forever if detection
> is not completed. Does the "i" have a timeout value?

No, there is no timeout. "i" was there to tell me how long it takes, but
I can use it to check for a timeout. I will fix this.
> Some indents are actually several spaces instead of tab. Please note
> this.

Thanks for telling me. I will clean up and resend. Some indents have
tabs and spaces to have exact indentation at the beginning of the
parentheses.

Regards,
Carl-Daniel
> Zheng
>
> -----Original Message-----
> From: Carl-Daniel Hailfinger

[mailto:[e-mail]]
> Sent: Tuesday, December 23, 2008 10:37 AM
> To: Bao, Zheng
> Cc: Coreboot
> Subject: Re: [PATCH] Fix SB600 SATA and add support for port 2-4
>
> Hi Zheng,
>
> On 23.12.2008 03:26, Bao, Zheng wrote:
> > Reviewed-by: zheng bao
> >
> > It is tested well on dbm690t and pistachio.
>
> Thank you for the fast review!
>
> Our commit process is a bit difficult because it needs an Acked-by:
> statement. (The subversion repository will reject a commit without
> Acked-by.)
>
> Everybody who understands the code is allowed to ack a patch. You
> understand that piece of code very well, so could you please also add

an
> Acked-by statement? Thanks!
>
> Regards,
> Carl-Daniel
> > -----Original Message-----
> > From: Carl-Daniel Hailfinger
>
> [mailto:[e-mail]]
> > Sent: Tuesday, December 23, 2008 10:01 AM
> > To: Coreboot
> > Cc: Bao, Zheng
> > Subject: [PATCH] Fix SB600 SATA and add support for port 2-4
> >
> > The SB600 RPR documentation does not mention what to do if

SATA_BAR0+6
> > is no longer 0xA0 or 0xB0. It simply assumes that will never happen.
> > My 500 GB Seagate Barracuda ST3500820AS triggers that corner case on
>
> the
> > first init after poweron.
> > The current code hangs forever with my drive. Fix this by rerunning
>
> the
> > init sequence after SATA_BAR0+6 is no longer 0xA0 or 0xB0.
> >
> > Add support for SATA port 2-4 (Primary Slave, Secondary Master,
> > Secondary Slave).
> >
> > Activate and improve debug messages for SPEW log level.
> >
> > Fix some comments.
> >
> > New log messages look like this:
> > PCI: 00:12.0 init
> > sata_bar0=3020
> > sata_bar1=3060
> > sata_bar2=3030
> > sata_bar3=3070
> > sata_bar4=3000
> > sata_bar5=fc309000
> > SATA port 0 status = 23
> > 0x6=a0, 0x7=80
> > drive detection not yet completed, waiting...
> > 0x6=a0, 0x7=80
> > drive detection not yet completed, waiting...
> > [... 281 repetitions ...]
> > 0x6=0, 0x7=50
> > drive no longer selected after 2830 ms, retrying init
> > drive detection done after 10 ms
> > Primary Master device is ready
> > SATA port 1 status = 23
> > drive detection done after 10 ms
> > Primary Slave device is ready
> > SATA port 2 status = 0
> > No Secondary Master SATA drive on Slot2
> > SATA port 3 status = 0
> > No Secondary Slave SATA drive on Slot3
> >
> > Full log is attached.
> >
> > With this patch (and my other non-SATA fixups), my Asus M2A-VM boots
> > into Linux without problems.
> >
> > Signed-off-by: Carl-Daniel Hailfinger

[PATCH] Fix SB600 SATA and add support for port 2-4

[Carl-Daniel Hailfinger]

On 23.12.2008 08:10, Bao, Zheng wrote:
> Sorry, I tested it again and tried it on 4 ports. It only works on 1st
> and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
>
> My SATA drive is 250 GB Seagate Barracuda ST3250620NS.

Thanks for testing!
This is very strange. It seems the hardware does not follow the BDG. Did
you have 4 drives attached at the same time or was it 1 drive attached
to another port each time?

Could you please write your own implementation of 4-port setup so we can
compare this?

Regards,
Carl-Daniel

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

I just tested as the latter one you said. That is, I have only 1 sata
drive, attached to each port every time.

We have CIM code in assembly code. I don't know if I can provide it. I
will check it ASAP. But now it is Christmas Eve in our time zone. I have
to go out and have fun. :)

Zheng

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Wednesday, December 24, 2008 4:00 PM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port
2-4

On 23.12.2008 08:10, Bao, Zheng wrote:
> Sorry, I tested it again and tried it on 4 ports. It only works on 1st
> and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
>
> My SATA drive is 250 GB Seagate Barracuda ST3250620NS.

Thanks for testing!
This is very strange. It seems the hardware does not follow the BDG. Did
you have 4 drives attached at the same time or was it 1 drive attached
to another port each time?

Could you please write your own implementation of 4-port setup so we can
compare this?

Regards,
Carl-Daniel

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

What you said reminded me to connect more drives when the board boots.
Yes, it changed. If 1st and 2nd ports are both connected, they can both
be correctly detected and initialized. Then I unplugged the drive on 1st
post and didn't change the BIOS, the 2nd ports couldn't be initialized
as it was. I check our asm code and can't find any tip.

Zheng

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Wednesday, December 24, 2008 4:00 PM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port
2-4

On 23.12.2008 08:10, Bao, Zheng wrote:
> Sorry, I tested it again and tried it on 4 ports. It only works on 1st
> and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
> driver no longer selected after 10ms, retrying init
>
> My SATA drive is 250 GB Seagate Barracuda ST3250620NS.

Thanks for testing!
This is very strange. It seems the hardware does not follow the BDG. Did
you have 4 drives attached at the same time or was it 1 drive attached
to another port each time?

Could you please write your own implementation of 4-port setup so we can
compare this?

Regards,
Carl-Daniel

[PATCH] Fix SB600 SATA and add support for port 2-4

[Carl-Daniel Hailfinger]

Hi Zheng,

On 25.12.2008 05:10, Bao, Zheng wrote:
> On 24.12.2008 16:00, Carl-Daniel Hailfinger wrote:
> > On 23.12.2008 08:10, Bao, Zheng wrote:
> > > Sorry, I tested it again and tried it on 4 ports. It only works on 1st
> > > and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> > > driver no longer selected after 10ms, retrying init
> > > driver no longer selected after 10ms, retrying init
> > >
> > > My SATA drive is 250 GB Seagate Barracuda ST3250620NS.
> >
> > Thanks for testing!
> > This is very strange. It seems the hardware does not follow the BDG. Did
> > you have 4 drives attached at the same time or was it 1 drive attached
> > to another port each time?
>
> What you said reminded me to connect more drives when the board boots.
> Yes, it changed. If 1st and 2nd ports are both connected, they can both
> be correctly detected and initialized. Then I unplugged the drive on 1st
> post and didn't change the BIOS, the 2nd ports couldn't be initialized
> as it was. I check our asm code and can't find any tip.

I have modified the code to avoid the endless loop and print better
diagnostic messages.
It will work if only 1st or if 1st and 2nd port are connected. If only
2nd is connected and the hardware acts strange, it will print an error
message and continue.
I have not tested 3rd and 4th port.

Can you please test if this works for you?

Signed-off-by: Carl-Daniel Hailfinger

Index: LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
===================================================================
--- LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c	(Revision 3844)
+++ LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c	(Arbeitskopie)
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2008 Advanced Micro Devices, Inc.
+ * Copyright (C) 2008 Carl-Daniel Hailfinger
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -26,6 +27,28 @@
 #include <arch/io.h>
 #include "sb600.h"

+int sata_drive_detect(int portnum, u16 iobar)
+{
+	u8 byte, byte2;
+	int i = 0;
+	outb(0xA0 + 0x10 * (portnum % 2), iobar + 0x6);
+	while (byte = inb(iobar + 0x6),
+	       byte2 = inb(iobar + 0x7),
+	       (byte != (0xA0 + 0x10 * (portnum % 2))) ||
+	       ((byte2 & 0x88) != 0)) {
+		printk_spew("0x6=%x, 0x7=%x\n", byte, byte2);
+		if (byte != (0xA0 + 0x10 * (portnum % 2))) {
+			printk_debug("drive no longer selected after %i ms, retrying init\n", i * 10);
+			return 1;
+		} else
+			printk_spew("drive detection not yet completed, waiting...\n");
+		mdelay(10);
+		i++;
+	}
+	printk_spew("drive detection done after %i ms\n", i * 10);
+	return 0;
+}
+
 static void sata_init(struct device *dev)
 {
    u8 byte;
@@ -33,6 +56,7 @@
    u32 dword;
    u8 *sata_bar5;
    u16 sata_bar0, sata_bar1, sata_bar2, sata_bar3, sata_bar4;
+	int i, j;

struct southbridge_ati_sb600_config *conf;
    conf = dev->chip_info;
@@ -62,12 +86,12 @@
    sata_bar3 = pci_read_config16(dev, 0x1C) & ~0x7;
    sata_bar4 = pci_read_config16(dev, 0x20) & ~0x7;

-	/* printk_debug("sata_bar0=%x\n", sata_bar0); */	/* 3030 */
-	/* printk_debug("sata_bar1=%x\n", sata_bar1); */	/* 3070 */
-	/* printk_debug("sata_bar2=%x\n", sata_bar2); */	/* 3040 */
-	/* printk_debug("sata_bar3=%x\n", sata_bar3); */	/* 3080 */
-	/* printk_debug("sata_bar4=%x\n", sata_bar4); */	/* 3000 */
-	/* printk_debug("sata_bar5=%x\n", sata_bar5); */	/* e0309000 */
+	printk_spew("sata_bar0=%x\n", sata_bar0);	/* 3030 */
+	printk_spew("sata_bar1=%x\n", sata_bar1);	/* 3070 */
+	printk_spew("sata_bar2=%x\n", sata_bar2);	/* 3040 */
+	printk_spew("sata_bar3=%x\n", sata_bar3);	/* 3080 */
+	printk_spew("sata_bar4=%x\n", sata_bar4);	/* 3000 */
+	printk_spew("sata_bar5=%x\n", sata_bar5);	/* e0309000 */

/* Program the 2C to 0x43801002 */
    dword = 0x43801002;
@@ -131,31 +155,36 @@
    dword |= 1 << 25;
    pci_write_config32(dev, 0x40, dword);

-	/* Enable the I/O ,MM ,BusMaster access for SATA */
+	/* Enable the I/O, MM, BusMaster access for SATA */
    byte = pci_read_config8(dev, 0x4);
    byte |= 7 << 0;
    pci_write_config8(dev, 0x4, byte);

-	/* RPR6.6 SATA drive detection. Currently we detect Primary Master Device only */
-	/* Use BAR5+0x1A8,BAR0+0x6 for Primary Slave */
-	/* Use BAR5+0x228,BAR0+0x6 for Secondary Master */
-	/* Use BAR5+0x2A8,BAR0+0x6 for Secondary Slave */
+	/* RPR6.6 SATA drive detection. */
+	/* Use BAR5+0x128,BAR0 for Primary Slave */
+	/* Use BAR5+0x1A8,BAR0 for Primary Slave */
+	/* Use BAR5+0x228,BAR2 for Secondary Master */
+	/* Use BAR5+0x2A8,BAR2 for Secondary Slave */

-	byte = readb(sata_bar5 + 0x128);
-	/* printk_debug("byte=%x\n", byte); */
-	byte &= 0xF;
-	if (byte == 0x3) {
-		outb(0xA0, sata_bar0 + 0x6);
-		while ((inb(sata_bar0 + 0x6) != 0xA0)
-		       || ((inb(sata_bar0 + 0x7) & 0x88) != 0)) {
-			mdelay(10);
-			/* printk_debug("0x6=%x,0x7=%x\n", inb(sata_bar0 + 0x6),
-			   inb(sata_bar0 + 0x7)); */
-			printk_debug("drive detection fail,trying...\n");
+	for (i = 0; i < 4; i++) {
+		byte = readb(sata_bar5 + 0x128 + 0x80 * i);
+		printk_spew("SATA port %i status = %x\n", i, byte);
+		byte &= 0xF;
+		if (byte == 0x3) {
+			for (j = 0; j < 10; j++) {
+				if (!sata_drive_detect(i, ((i / 2) == 0) ? sata_bar0 : sata_bar2))
+					break;
+			}
+			printk_debug("%s %s device is %sready after %i tries\n",
+				     (i / 2) ? "Secondary" : "Primary",
+				     (i % 2 ) ? "Slave" : "Master",
+				     (j == 10) ? "not " : "",
+				     (j == 10) ? j : j + 1);
+		} else {
+			printk_debug("No %s %s SATA drive on Slot%i\n",
+				     (i / 2) ? "Secondary" : "Primary",
+				     (i % 2 ) ? "Slave" : "Master", i);
    	}
-		printk_debug("Primary master device is ready\n");
-	} else {
-		printk_debug("No Primary master SATA drive on Slot0\n");
    }

/* Below is CIM InitSataLateFar */

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

It seems to be a resolution before we can finally get the reason why
hardware behaves strange.

Please replace the multiple spaces with tab before checking in.

Acked-by: Zheng Bao

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Monday, December 29, 2008 10:48 AM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port
2-4

Hi Zheng,

On 25.12.2008 05:10, Bao, Zheng wrote:
> On 24.12.2008 16:00, Carl-Daniel Hailfinger wrote:
> > On 23.12.2008 08:10, Bao, Zheng wrote:
> > > Sorry, I tested it again and tried it on 4 ports. It only works on

1st
> > > and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> > > driver no longer selected after 10ms, retrying init
> > > driver no longer selected after 10ms, retrying init
> > >
> > > My SATA drive is 250 GB Seagate Barracuda ST3250620NS.
> >
> > Thanks for testing!
> > This is very strange. It seems the hardware does not follow the BDG.

Did
> > you have 4 drives attached at the same time or was it 1 drive

attached
> > to another port each time?
>
> What you said reminded me to connect more drives when the board boots.
> Yes, it changed. If 1st and 2nd ports are both connected, they can

both
> be correctly detected and initialized. Then I unplugged the drive on

1st
> post and didn't change the BIOS, the 2nd ports couldn't be initialized
> as it was. I check our asm code and can't find any tip.

I have modified the code to avoid the endless loop and print better
diagnostic messages.
It will work if only 1st or if 1st and 2nd port are connected. If only
2nd is connected and the hardware acts strange, it will print an error
message and continue.
I have not tested 3rd and 4th port.

Can you please test if this works for you?

Signed-off-by: Carl-Daniel Hailfinger

Index: LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
===================================================================
--- LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
(Revision 3844)
+++ LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
(Arbeitskopie)
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2008 Advanced Micro Devices, Inc.
+ * Copyright (C) 2008 Carl-Daniel Hailfinger
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -26,6 +27,28 @@
 #include <arch/io.h>
 #include "sb600.h"

+int sata_drive_detect(int portnum, u16 iobar)
+{
+	u8 byte, byte2;
+	int i = 0;
+	outb(0xA0 + 0x10 * (portnum % 2), iobar + 0x6);
+	while (byte = inb(iobar + 0x6),
+	       byte2 = inb(iobar + 0x7),
+	       (byte != (0xA0 + 0x10 * (portnum % 2))) ||
+	       ((byte2 & 0x88) != 0)) {
+		printk_spew("0x6=%x, 0x7=%x\n", byte, byte2);
+		if (byte != (0xA0 + 0x10 * (portnum % 2))) {
+			printk_debug("drive no longer selected after %i
ms, retrying init\n", i * 10);
+			return 1;
+		} else
+			printk_spew("drive detection not yet completed,
waiting...\n");
+		mdelay(10);
+		i++;
+	}
+	printk_spew("drive detection done after %i ms\n", i * 10);
+	return 0;
+}
+
 static void sata_init(struct device *dev)
 {
    u8 byte;
@@ -33,6 +56,7 @@
    u32 dword;
    u8 *sata_bar5;
    u16 sata_bar0, sata_bar1, sata_bar2, sata_bar3, sata_bar4;
+	int i, j;

struct southbridge_ati_sb600_config *conf;
    conf = dev->chip_info;
@@ -62,12 +86,12 @@
    sata_bar3 = pci_read_config16(dev, 0x1C) & ~0x7;
    sata_bar4 = pci_read_config16(dev, 0x20) & ~0x7;

-	/* printk_debug("sata_bar0=%x\n", sata_bar0); */	/* 3030
*/
-	/* printk_debug("sata_bar1=%x\n", sata_bar1); */	/* 3070
*/
-	/* printk_debug("sata_bar2=%x\n", sata_bar2); */	/* 3040
*/
-	/* printk_debug("sata_bar3=%x\n", sata_bar3); */	/* 3080
*/
-	/* printk_debug("sata_bar4=%x\n", sata_bar4); */	/* 3000
*/
-	/* printk_debug("sata_bar5=%x\n", sata_bar5); */	/*
e0309000 */
+	printk_spew("sata_bar0=%x\n", sata_bar0);	/* 3030 */
+	printk_spew("sata_bar1=%x\n", sata_bar1);	/* 3070 */
+	printk_spew("sata_bar2=%x\n", sata_bar2);	/* 3040 */
+	printk_spew("sata_bar3=%x\n", sata_bar3);	/* 3080 */
+	printk_spew("sata_bar4=%x\n", sata_bar4);	/* 3000 */
+	printk_spew("sata_bar5=%x\n", sata_bar5);	/* e0309000 */

/* Program the 2C to 0x43801002 */
    dword = 0x43801002;
@@ -131,31 +155,36 @@
    dword |= 1 << 25;
    pci_write_config32(dev, 0x40, dword);

-	/* Enable the I/O ,MM ,BusMaster access for SATA */
+	/* Enable the I/O, MM, BusMaster access for SATA */
    byte = pci_read_config8(dev, 0x4);
    byte |= 7 << 0;
    pci_write_config8(dev, 0x4, byte);

-	/* RPR6.6 SATA drive detection. Currently we detect Primary
Master Device only */
-	/* Use BAR5+0x1A8,BAR0+0x6 for Primary Slave */
-	/* Use BAR5+0x228,BAR0+0x6 for Secondary Master */
-	/* Use BAR5+0x2A8,BAR0+0x6 for Secondary Slave */
+	/* RPR6.6 SATA drive detection. */
+	/* Use BAR5+0x128,BAR0 for Primary Slave */
+	/* Use BAR5+0x1A8,BAR0 for Primary Slave */
+	/* Use BAR5+0x228,BAR2 for Secondary Master */
+	/* Use BAR5+0x2A8,BAR2 for Secondary Slave */

-	byte = readb(sata_bar5 + 0x128);
-	/* printk_debug("byte=%x\n", byte); */
-	byte &= 0xF;
-	if (byte == 0x3) {
-		outb(0xA0, sata_bar0 + 0x6);
-		while ((inb(sata_bar0 + 0x6) != 0xA0)
-		       || ((inb(sata_bar0 + 0x7) & 0x88) != 0)) {
-			mdelay(10);
-			/* printk_debug("0x6=%x,0x7=%x\n", inb(sata_bar0
+ 0x6),
-			   inb(sata_bar0 + 0x7)); */
-			printk_debug("drive detection
fail,trying...\n");
+	for (i = 0; i < 4; i++) {
+		byte = readb(sata_bar5 + 0x128 + 0x80 * i);
+		printk_spew("SATA port %i status = %x\n", i, byte);
+		byte &= 0xF;
+		if (byte == 0x3) {
+			for (j = 0; j < 10; j++) {
+				if (!sata_drive_detect(i, ((i / 2) == 0)
? sata_bar0 : sata_bar2))
+					break;
+			}
+			printk_debug("%s %s device is %sready after %i
tries\n",
+				     (i / 2) ? "Secondary" : "Primary",
+				     (i % 2 ) ? "Slave" : "Master",
+				     (j == 10) ? "not " : "",
+				     (j == 10) ? j : j + 1);
+		} else {
+			printk_debug("No %s %s SATA drive on Slot%i\n",
+				     (i / 2) ? "Secondary" : "Primary",
+				     (i % 2 ) ? "Slave" : "Master", i);
    	}
-		printk_debug("Primary master device is ready\n");
-	} else {
-		printk_debug("No Primary master SATA drive on Slot0\n");
    }

/* Below is CIM InitSataLateFar */

[PATCH] Fix SB600 SATA and add support for port 2-4

[Carl-Daniel Hailfinger]

On 29.12.2008 07:28, Bao, Zheng wrote:
> It seems to be a resolution before we can finally get the reason why
> hardware behaves strange.
>
> Please replace the multiple spaces with tab before checking in.

Sure.
> Acked-by: Zheng Bao

Thanks, committed in revision 3845.

Regards,
Carl-Daniel
> -----Original Message-----
> From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
> Sent: Monday, December 29, 2008 10:48 AM
> To: Bao, Zheng
> Cc: Coreboot
> Subject: Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port
> 2-4
>
> Hi Zheng,
>
> On 25.12.2008 05:10, Bao, Zheng wrote:
> > On 24.12.2008 16:00, Carl-Daniel Hailfinger wrote:
> > > On 23.12.2008 08:10, Bao, Zheng wrote:
> > > > Sorry, I tested it again and tried it on 4 ports. It only works on 1st
> > > > and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> > > > driver no longer selected after 10ms, retrying init
> > > > driver no longer selected after 10ms, retrying init
> > > >
> > > > My SATA drive is 250 GB Seagate Barracuda ST3250620NS.
> > >
> > > Thanks for testing!
> > > This is very strange. It seems the hardware does not follow the BDG.
> > > Did you have 4 drives attached at the same time or was it 1 drive attached
> > > to another port each time?
> >
> > What you said reminded me to connect more drives when the board boots.
> > Yes, it changed. If 1st and 2nd ports are both connected, they can
> > both be
> > correctly detected and initialized. Then I unplugged the drive on 1st
> > post and didn't change the BIOS, the 2nd ports couldn't be initialized
> > as it was. I check our asm code and can't find any tip.
>
> I have modified the code to avoid the endless loop and print better
> diagnostic messages.
> It will work if only 1st or if 1st and 2nd port are connected. If only
> 2nd is connected and the hardware acts strange, it will print an error
> message and continue.
> I have not tested 3rd and 4th port.
>
> Can you please test if this works for you?
>
> Signed-off-by: Carl-Daniel Hailfinger

[PATCH] Fix SB600 SATA and add support for port 2-4

[Bao, Zheng]

I tried with proprietary BIOS and access BAR0+6 with debugging tool.
Bar0+6 are always 0x7F like it is when coreboot boots. I believe that is
some kind of hardware misleading feature.

Actually, the filo can always found the file system on each SATA port. I
think that is it. I hope.

Zheng

-----Original Message-----
From: Carl-Daniel Hailfinger [mailto:[e-mail]] 
Sent: Monday, December 29, 2008 10:48 AM
To: Bao, Zheng
Cc: Coreboot
Subject: Re: [coreboot] [PATCH] Fix SB600 SATA and add support for port
2-4

Hi Zheng,

On 25.12.2008 05:10, Bao, Zheng wrote:
> On 24.12.2008 16:00, Carl-Daniel Hailfinger wrote:
> > On 23.12.2008 08:10, Bao, Zheng wrote:
> > > Sorry, I tested it again and tried it on 4 ports. It only works on

1st
> > > and 4th ports, while doesn't work on 2nd and 3rd ports. It loops at
> > > driver no longer selected after 10ms, retrying init
> > > driver no longer selected after 10ms, retrying init
> > >
> > > My SATA drive is 250 GB Seagate Barracuda ST3250620NS.
> >
> > Thanks for testing!
> > This is very strange. It seems the hardware does not follow the BDG.

Did
> > you have 4 drives attached at the same time or was it 1 drive

attached
> > to another port each time?
>
> What you said reminded me to connect more drives when the board boots.
> Yes, it changed. If 1st and 2nd ports are both connected, they can

both
> be correctly detected and initialized. Then I unplugged the drive on

1st
> post and didn't change the BIOS, the 2nd ports couldn't be initialized
> as it was. I check our asm code and can't find any tip.

I have modified the code to avoid the endless loop and print better
diagnostic messages.
It will work if only 1st or if 1st and 2nd port are connected. If only
2nd is connected and the hardware acts strange, it will print an error
message and continue.
I have not tested 3rd and 4th port.

Can you please test if this works for you?

Signed-off-by: Carl-Daniel Hailfinger

Index: LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
===================================================================
--- LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
(Revision 3844)
+++ LinuxBIOSv2-asus_m2a-vm/src/southbridge/amd/sb600/sb600_sata.c
(Arbeitskopie)
@@ -2,6 +2,7 @@
  * This file is part of the coreboot project.
  *
  * Copyright (C) 2008 Advanced Micro Devices, Inc.
+ * Copyright (C) 2008 Carl-Daniel Hailfinger
  *
  * This program is free software; you can redistribute it and/or modify
  * it under the terms of the GNU General Public License as published by
@@ -26,6 +27,28 @@
 #include <arch/io.h>
 #include "sb600.h"

+int sata_drive_detect(int portnum, u16 iobar)
+{
+	u8 byte, byte2;
+	int i = 0;
+	outb(0xA0 + 0x10 * (portnum % 2), iobar + 0x6);
+	while (byte = inb(iobar + 0x6),
+	       byte2 = inb(iobar + 0x7),
+	       (byte != (0xA0 + 0x10 * (portnum % 2))) ||
+	       ((byte2 & 0x88) != 0)) {
+		printk_spew("0x6=%x, 0x7=%x\n", byte, byte2);
+		if (byte != (0xA0 + 0x10 * (portnum % 2))) {
+			printk_debug("drive no longer selected after %i
ms, retrying init\n", i * 10);
+			return 1;
+		} else
+			printk_spew("drive detection not yet completed,
waiting...\n");
+		mdelay(10);
+		i++;
+	}
+	printk_spew("drive detection done after %i ms\n", i * 10);
+	return 0;
+}
+
 static void sata_init(struct device *dev)
 {
    u8 byte;
@@ -33,6 +56,7 @@
    u32 dword;
    u8 *sata_bar5;
    u16 sata_bar0, sata_bar1, sata_bar2, sata_bar3, sata_bar4;
+	int i, j;

struct southbridge_ati_sb600_config *conf;
    conf = dev->chip_info;
@@ -62,12 +86,12 @@
    sata_bar3 = pci_read_config16(dev, 0x1C) & ~0x7;
    sata_bar4 = pci_read_config16(dev, 0x20) & ~0x7;

-	/* printk_debug("sata_bar0=%x\n", sata_bar0); */	/* 3030
*/
-	/* printk_debug("sata_bar1=%x\n", sata_bar1); */	/* 3070
*/
-	/* printk_debug("sata_bar2=%x\n", sata_bar2); */	/* 3040
*/
-	/* printk_debug("sata_bar3=%x\n", sata_bar3); */	/* 3080
*/
-	/* printk_debug("sata_bar4=%x\n", sata_bar4); */	/* 3000
*/
-	/* printk_debug("sata_bar5=%x\n", sata_bar5); */	/*
e0309000 */
+	printk_spew("sata_bar0=%x\n", sata_bar0);	/* 3030 */
+	printk_spew("sata_bar1=%x\n", sata_bar1);	/* 3070 */
+	printk_spew("sata_bar2=%x\n", sata_bar2);	/* 3040 */
+	printk_spew("sata_bar3=%x\n", sata_bar3);	/* 3080 */
+	printk_spew("sata_bar4=%x\n", sata_bar4);	/* 3000 */
+	printk_spew("sata_bar5=%x\n", sata_bar5);	/* e0309000 */

/* Program the 2C to 0x43801002 */
    dword = 0x43801002;
@@ -131,31 +155,36 @@
    dword |= 1 << 25;
    pci_write_config32(dev, 0x40, dword);

-	/* Enable the I/O ,MM ,BusMaster access for SATA */
+	/* Enable the I/O, MM, BusMaster access for SATA */
    byte = pci_read_config8(dev, 0x4);
    byte |= 7 << 0;
    pci_write_config8(dev, 0x4, byte);

-	/* RPR6.6 SATA drive detection. Currently we detect Primary
Master Device only */
-	/* Use BAR5+0x1A8,BAR0+0x6 for Primary Slave */
-	/* Use BAR5+0x228,BAR0+0x6 for Secondary Master */
-	/* Use BAR5+0x2A8,BAR0+0x6 for Secondary Slave */
+	/* RPR6.6 SATA drive detection. */
+	/* Use BAR5+0x128,BAR0 for Primary Slave */
+	/* Use BAR5+0x1A8,BAR0 for Primary Slave */
+	/* Use BAR5+0x228,BAR2 for Secondary Master */
+	/* Use BAR5+0x2A8,BAR2 for Secondary Slave */

-	byte = readb(sata_bar5 + 0x128);
-	/* printk_debug("byte=%x\n", byte); */
-	byte &= 0xF;
-	if (byte == 0x3) {
-		outb(0xA0, sata_bar0 + 0x6);
-		while ((inb(sata_bar0 + 0x6) != 0xA0)
-		       || ((inb(sata_bar0 + 0x7) & 0x88) != 0)) {
-			mdelay(10);
-			/* printk_debug("0x6=%x,0x7=%x\n", inb(sata_bar0
+ 0x6),
-			   inb(sata_bar0 + 0x7)); */
-			printk_debug("drive detection
fail,trying...\n");
+	for (i = 0; i < 4; i++) {
+		byte = readb(sata_bar5 + 0x128 + 0x80 * i);
+		printk_spew("SATA port %i status = %x\n", i, byte);
+		byte &= 0xF;
+		if (byte == 0x3) {
+			for (j = 0; j < 10; j++) {
+				if (!sata_drive_detect(i, ((i / 2) == 0)
? sata_bar0 : sata_bar2))
+					break;
+			}
+			printk_debug("%s %s device is %sready after %i
tries\n",
+				     (i / 2) ? "Secondary" : "Primary",
+				     (i % 2 ) ? "Slave" : "Master",
+				     (j == 10) ? "not " : "",
+				     (j == 10) ? j : j + 1);
+		} else {
+			printk_debug("No %s %s SATA drive on Slot%i\n",
+				     (i / 2) ? "Secondary" : "Primary",
+				     (i % 2 ) ? "Slave" : "Master", i);
    	}
-		printk_debug("Primary master device is ready\n");
-	} else {
-		printk_debug("No Primary master SATA drive on Slot0\n");
    }

/* Below is CIM InitSataLateFar */

[PATCH] Fix SB600 SATA and add support for port 2-4

[Carl-Daniel Hailfinger]

On 29.12.2008 07:55, Bao, Zheng wrote:
> I tried with proprietary BIOS and access BAR0+6 with debugging tool.
> Bar0+6 are always 0x7F like it is when coreboot boots. I believe that is
> some kind of hardware misleading feature.
>
> Actually, the filo can always found the file system on each SATA port. I
> think that is it. I hope.

OK. I hope that is good enough. Could you tell the BDG writers at AMD
about this problem so they can mention it in the BDG with a workaround?
Thanks!

Regards,
Carl-Daniel