ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
Signed-off-by: [email protected]
---------
--- linux-2.6.19-rc6-git4/drivers/pci/quirks.c.orig 2006-11-23 19:45:49.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/pci/quirks.c 2006-11-23 19:34:23.000000000 +0800
@@ -795,6 +795,25 @@ static void __init quirk_mediagx_master(
}
}
DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_CYRIX, PCI_DEVICE_ID_CYRIX_PCI_MASTER, quirk_mediagx_master );
+
+#if defined(CONFIG_SATA_AHCI) || defined(CONFIG_SATA_AHCI_MODULE)
+static void __devinit quirk_sb600_sata(struct pci_dev *pdev)
+{
+ /* set sb600 sata to ahci mode */
+ if ((pdev->class >> 8) == PCI_CLASS_STORAGE_IDE) {
+ u8 tmp;
+
+ pci_read_config_byte(pdev, 0x40, &tmp);
+ pci_write_config_byte(pdev, 0x40, tmp|1);
+ pci_write_config_byte(pdev, 0x9, 1);
+ pci_write_config_byte(pdev, 0xa, 6);
+ pci_write_config_byte(pdev, 0x40, tmp);
+
+ pdev->class = 0x010601;
+ }
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP600_SATA, quirk_sb600_sata);
+#endif
/*
* As per PCI spec, ignore base address registers 0-3 of the IDE controllers
-----Original Message-----
From: Conke Hu
Sent: Thursday, November 23, 2006 12:21 PM
To: [email protected]; '[email protected]'; 'Andrew Morton'; Jeff Garzik
Subject: [PATCH] Add IDE mode support for SB600 SATA
ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
Signed-off-by: [email protected]
---------
--- linux-2.6.19-rc6-git4/drivers/pci/quirks.c.orig 2006-11-23 19:45:49.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/pci/quirks.c 2006-11-23 19:34:23.000000000 +0800
@@ -795,6 +795,25 @@ static void __init quirk_mediagx_master(
}
}
DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_CYRIX, PCI_DEVICE_ID_CYRIX_PCI_MASTER, quirk_mediagx_master );
+
+#if defined(CONFIG_SATA_AHCI) || defined(CONFIG_SATA_AHCI_MODULE)
+static void __devinit quirk_sb600_sata(struct pci_dev *pdev) {
+ /* set sb600 sata to ahci mode */
+ if ((pdev->class >> 8) == PCI_CLASS_STORAGE_IDE) {
+ u8 tmp;
+
+ pci_read_config_byte(pdev, 0x40, &tmp);
+ pci_write_config_byte(pdev, 0x40, tmp|1);
+ pci_write_config_byte(pdev, 0x9, 1);
+ pci_write_config_byte(pdev, 0xa, 6);
+ pci_write_config_byte(pdev, 0x40, tmp);
+
+ pdev->class = 0x010601;
+ }
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI,
+PCI_DEVICE_ID_ATI_IXP600_SATA, quirk_sb600_sata); #endif
/*
* As per PCI spec, ignore base address registers 0-3 of the IDE controllers
---------------------
This is the re-written patch, if still any unreasonable, please feel free to contact me. Thanks!
Conke
On Thu, 2006-11-23 at 12:20 +0800, Conke Hu wrote:
> ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
is this really the right thing? You're overriding a user chosen
configuration here.... while that might be justifiable.. it's probably a
good idea to at least provide a safety-valve for this one. The user
might have made that selection very deliberately.
--
if you want to mail me at work (you don't), use arjan (at) linux.intel.com
Test the interaction between Linux and your BIOS via http://www.linuxfirmwarekit.org
-----Original Message-----
From: Arjan van de Ven [mailto:[email protected]]
Sent: Thursday, November 23, 2006 4:06 PM
To: Conke Hu
Cc: [email protected]; [email protected]; Andrew Morton; Jeff Garzik
Subject: Re: [PATCH] Add IDE mode support for SB600 SATA
On Thu, 2006-11-23 at 12:20 +0800, Conke Hu wrote:
> ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
is this really the right thing? You're overriding a user chosen configuration here.... while that might be justifiable.. it's probably a good idea to at least provide a safety-valve for this one. The user might have made that selection very deliberately.
--
if you want to mail me at work (you don't), use arjan (at) linux.intel.com Test the interaction between Linux and your BIOS via http://www.linuxfirmwarekit.org
-------------------------------
Hi Arjan,
In my preivious patch(with the same email title "[PATCH] Add IDE mode support for SB600 SATA"), I provided a config option and kernel users (the "users" here includes kernel developers and linux fans) can use the sata controller as IDE or AHCI, but Andrew did not think that was good idea.
[Quote from Andrew]
I doubt if it's appropriate to do all this via ifdefs. Users don't compile their kernels - others compile them for the users. We need the one kernel
binary to support both modes. Possible?
Andrew means only one mode (i.e. ahci) support is enough, so I've re-writen this patch according to Alan's adavice, see bellow:
[Quote from Alan Cox]
That seems fine to me. I would have thought putting the code you have into the quirks.c file as you proposed was the better way to do this, but with the addition of the
#if defined (CONFIG_ATA_AHCI) || defined(CONFIG_ATA_AHCI_MODULE)
#endif
around it
Hi Alan,
Today I've sent out 2 patches about the same issue. one patch/solution is providing a config option to support both IDE driver and AHCI driver; another is using ahci driver for all modes. The two patches respectively corresond to two mails with the same title "[PATCH] Add IDE mode support for SB600 SATA", please take a look at both patches for more details.
Which patch/solution do you think is acceptable? Or is there anything that needs to be improved?
Thank you!
Conke
-----Original Message-----
From: Arjan van de Ven [mailto:[email protected]]
Sent: Thursday, November 23, 2006 4:06 PM
To: Conke Hu
Cc: [email protected]; [email protected]; Andrew Morton; Jeff Garzik
Subject: Re: [PATCH] Add IDE mode support for SB600 SATA
On Thu, 2006-11-23 at 12:20 +0800, Conke Hu wrote:
> ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
is this really the right thing? You're overriding a user chosen configuration here.... while that might be justifiable.. it's probably a good idea to at least provide a safety-valve for this one. The user might have made that selection very deliberately.
--
if you want to mail me at work (you don't), use arjan (at) linux.intel.com Test the interaction between Linux and your BIOS via http://www.linuxfirmwarekit.org
---------------------------------------------------------------------------------
Hi all,
For all your convenience, I paste 2 patches here, pls see following.
For the chip view, I think, either of them is reasonable, but kernel module maintainers maybe have more consideration, so if the module maintainers (Alan Cox?) have any advice, please let me know and I will modify my patch as soon as possible :) Thanks!
=============== solution 1 ==================
diff -Nur linux-2.6.19-rc6-git4.orig/drivers/ata/ahci.c linux-2.6.19-rc6-git4/drivers/ata/ahci.c
--- linux-2.6.19-rc6-git4.orig/drivers/ata/ahci.c 2006-11-23 13:36:52.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/ata/ahci.c 2006-11-23 13:50:13.000000000 +0800
@@ -323,7 +323,14 @@
{ PCI_VDEVICE(JMICRON, 0x2366), board_ahci }, /* JMicron JMB366 */
/* ATI */
+#ifdef CONFIG_SB600_AHCI_IDE
{ PCI_VDEVICE(ATI, 0x4380), board_ahci }, /* ATI SB600 non-raid */
+#else
+ { PCI_VENDOR_ID_ATI, 0x4380, PCI_ANY_ID, PCI_ANY_ID, PCI_CLASS_STORAGE_RAID<<8, 0xffff00,
+ board_ahci },
+ { PCI_VENDOR_ID_ATI, 0x4380, PCI_ANY_ID, PCI_ANY_ID, 0x010600, 0xffff00,
+ board_ahci },
+#endif
{ PCI_VDEVICE(ATI, 0x4381), board_ahci }, /* ATI SB600 raid */
/* VIA */
diff -Nur linux-2.6.19-rc6-git4.orig/drivers/ata/Kconfig linux-2.6.19-rc6-git4/drivers/ata/Kconfig
--- linux-2.6.19-rc6-git4.orig/drivers/ata/Kconfig 2006-11-23 13:36:52.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/ata/Kconfig 2006-11-23 13:37:50.000000000 +0800
@@ -27,6 +27,18 @@
If unsure, say N.
+config SB600_AHCI_IDE
+ bool "ATI SB600 IDE mode of SATA controller support"
+ depends on SATA_AHCI
+ default y
+ help
+ ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native
+ IDE, AHCI and RAID. If this option is enabled, the contoller
+ will run in AHCI mode and use ahci.ko as it's driver, otherwise
+ it will work as an IDE controller using atiixp.ko as it's driver.
+
+ If unsure, say y.
+
config SATA_SVW
tristate "ServerWorks Frodo / Apple K2 SATA support"
depends on PCI
diff -Nur linux-2.6.19-rc6-git4.orig/drivers/ide/pci/atiixp.c linux-2.6.19-rc6-git4/drivers/ide/pci/atiixp.c
--- linux-2.6.19-rc6-git4.orig/drivers/ide/pci/atiixp.c 2006-11-23 13:36:51.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/ide/pci/atiixp.c 2006-11-23 13:38:08.000000000 +0800
@@ -318,7 +318,8 @@
hwif->drives[0].autodma = hwif->autodma; }
-static void __devinit init_hwif_sb600_legacy(ide_hwif_t *hwif)
+#ifndef CONFIG_SB600_AHCI_IDE
+static void __devinit init_hwif_sb600_sata(ide_hwif_t *hwif)
{
hwif->atapi_dma = 1;
@@ -331,6 +332,7 @@
hwif->drives[0].autodma = hwif->autodma;
hwif->drives[1].autodma = hwif->autodma; }
+#endif
static ide_pci_device_t atiixp_pci_info[] __devinitdata = {
{ /* 0 */
@@ -340,13 +342,16 @@
.autodma = AUTODMA,
.enablebits = {{0x48,0x01,0x00}, {0x48,0x08,0x00}},
.bootable = ON_BOARD,
- },{ /* 1 */
- .name = "ATI SB600 SATA Legacy IDE",
- .init_hwif = init_hwif_sb600_legacy,
+ },
+#ifndef CONFIG_SB600_AHCI_IDE
+ { /* 1 */
+ .name = "ATI SB600 SATA (IDE mode)",
+ .init_hwif = init_hwif_sb600_sata,
.channels = 2,
.autodma = AUTODMA,
.bootable = ON_BOARD,
}
+#endif
};
/**
@@ -368,7 +373,9 @@
{ PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP300_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
{ PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP400_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
{ PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP600_IDE, PCI_ANY_ID, PCI_ANY_ID, 0, 0, 0},
- { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP600_SATA, PCI_ANY_ID, PCI_ANY_ID, (PCI_CLASS_STORAGE_IDE<<8)|0x8a, 0xffff05, 1},
+#ifndef CONFIG_SB600_AHCI_IDE
+ { PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP600_SATA, PCI_ANY_ID,
+PCI_ANY_ID, PCI_CLASS_STORAGE_IDE<<8, 0xffff00, 1}, #endif
{ 0, },
};
MODULE_DEVICE_TABLE(pci, atiixp_pci_tbl); diff -Nur linux-2.6.19-rc6-git4.orig/drivers/pci/quirks.c linux-2.6.19-rc6-git4/drivers/pci/quirks.c
--- linux-2.6.19-rc6-git4.orig/drivers/pci/quirks.c 2006-11-23 13:36:52.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/pci/quirks.c 2006-11-23 13:38:17.000000000 +0800
@@ -796,6 +796,25 @@
}
DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_CYRIX, PCI_DEVICE_ID_CYRIX_PCI_MASTER, quirk_mediagx_master );
+#ifdef CONFIG_SB600_AHCI_IDE
+static void __devinit quirk_sb600_sata(struct pci_dev *pdev) {
+ if ((pdev->class >> 8) == PCI_CLASS_STORAGE_IDE) {
+ u8 tmp;
+
+ pci_read_config_byte(pdev, 0x40, &tmp);
+ pci_write_config_byte(pdev, 0x40, tmp|1);
+ pci_write_config_byte(pdev, 0x9, 1);
+ pci_write_config_byte(pdev, 0xa, 6);
+ pci_write_config_byte(pdev, 0x40, tmp);
+
+ pdev->class = 0x010601;
+ }
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI,
+PCI_DEVICE_ID_ATI_IXP600_SATA, quirk_sb600_sata); #endif
+
+
/*
* As per PCI spec, ignore base address registers 0-3 of the IDE controllers
* running in Compatible mode (bits 0 and 2 in the ProgIf for primary and
===================== solution 2 ========================
--- linux-2.6.19-rc6-git4/drivers/pci/quirks.c.orig 2006-11-23 19:45:49.000000000 +0800
+++ linux-2.6.19-rc6-git4/drivers/pci/quirks.c 2006-11-23 19:34:23.000000000 +0800
@@ -795,6 +795,25 @@ static void __init quirk_mediagx_master(
}
}
DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_CYRIX, PCI_DEVICE_ID_CYRIX_PCI_MASTER, quirk_mediagx_master );
+
+#if defined(CONFIG_SATA_AHCI) || defined(CONFIG_SATA_AHCI_MODULE)
+static void __devinit quirk_sb600_sata(struct pci_dev *pdev) {
+ /* set sb600 sata to ahci mode */
+ if ((pdev->class >> 8) == PCI_CLASS_STORAGE_IDE) {
+ u8 tmp;
+
+ pci_read_config_byte(pdev, 0x40, &tmp);
+ pci_write_config_byte(pdev, 0x40, tmp|1);
+ pci_write_config_byte(pdev, 0x9, 1);
+ pci_write_config_byte(pdev, 0xa, 6);
+ pci_write_config_byte(pdev, 0x40, tmp);
+
+ pdev->class = 0x010601;
+ }
+}
+DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI,
+PCI_DEVICE_ID_ATI_IXP600_SATA, quirk_sb600_sata); #endif
/*
* As per PCI spec, ignore base address registers 0-3 of the IDE controllers
> is this really the right thing? You're overriding a user chosen
> configuration here.... while that might be justifiable.. it's probably a
> good idea to at least provide a safety-valve for this one. The user
> might have made that selection very deliberately.
Its what we do for other similar cases and I think its the right thing to
do in this situation. One reason for this is that with multi-boot boxes
you have to set the BIOS option to the dumbest one unless the smart OS's
reconfigure the device.
Alan
On Thu, 23 Nov 2006 12:20:37 +0800
"Conke Hu" <[email protected]> wrote:
>
> ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
>
> Signed-off-by: [email protected]
Acked-by: Alan Cox <[email protected]>
On Thu, 2006-11-23 at 11:22 +0000, Alan wrote:
> > is this really the right thing? You're overriding a user chosen
> > configuration here.... while that might be justifiable.. it's probably a
> > good idea to at least provide a safety-valve for this one. The user
> > might have made that selection very deliberately.
>
> Its what we do for other similar cases and I think its the right thing to
> do in this situation. One reason for this is that with multi-boot boxes
> you have to set the BIOS option to the dumbest one unless the smart OS's
> reconfigure the device.
while I can appreciate that.. it does assume things like SMM not
assuming things about it; and on resume.. has to happen as again/well :)
--
if you want to mail me at work (you don't), use arjan (at) linux.intel.com
Test the interaction between Linux and your BIOS via http://www.linuxfirmwarekit.org
Conke Hu wrote:
> ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE, AHCI and RAID. Legacy/Native IDE mode is designed for compatibility with some old OS without AHCI driver but looses SATAII/AHCI features such as NCQ. This patch will make SB600 SATA run in AHCI mode even if it was set as IDE mode by system BIOS.
>
> Signed-off-by: [email protected]
> ---------
> --- linux-2.6.19-rc6-git4/drivers/pci/quirks.c.orig 2006-11-23 19:45:49.000000000 +0800
> +++ linux-2.6.19-rc6-git4/drivers/pci/quirks.c 2006-11-23 19:34:23.000000000 +0800
> @@ -795,6 +795,25 @@ static void __init quirk_mediagx_master(
> }
> }
> DECLARE_PCI_FIXUP_FINAL(PCI_VENDOR_ID_CYRIX, PCI_DEVICE_ID_CYRIX_PCI_MASTER, quirk_mediagx_master );
> +
> +#if defined(CONFIG_SATA_AHCI) || defined(CONFIG_SATA_AHCI_MODULE)
> +static void __devinit quirk_sb600_sata(struct pci_dev *pdev)
> +{
> + /* set sb600 sata to ahci mode */
> + if ((pdev->class >> 8) == PCI_CLASS_STORAGE_IDE) {
> + u8 tmp;
> +
> + pci_read_config_byte(pdev, 0x40, &tmp);
> + pci_write_config_byte(pdev, 0x40, tmp|1);
> + pci_write_config_byte(pdev, 0x9, 1);
> + pci_write_config_byte(pdev, 0xa, 6);
> + pci_write_config_byte(pdev, 0x40, tmp);
> +
Two trailing tabs in the above line. Please remove those.
> + pdev->class = 0x010601;
> + }
> +}
> +DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_ATI, PCI_DEVICE_ID_ATI_IXP600_SATA, quirk_sb600_sata);
> +#endif
>
> /*
> * As per PCI spec, ignore base address registers 0-3 of the IDE controllers
Other than that, Acked-by: Tejun Heo <[email protected]>
--
tejun
Tejun Heo wrote:
> Conke Hu wrote:
>> ATI SB600 SATA controller supports 4 modes: Legacy IDE, Native IDE,
>> AHCI and RAID. Legacy/Native IDE mode is designed for compatibility
>> with some old OS without AHCI driver but looses SATAII/AHCI features
>> such as NCQ. This patch will make SB600 SATA run in AHCI mode even if
>> it was set as IDE mode by system BIOS.
[--snip--]
> Other than that, Acked-by: Tejun Heo <[email protected]>
At the second thought, I think this should be done in
ahci_init_controller().
* Unlike Jmicron's case, this doesn't affect PCI bus scan. Actually, it
does change class code but that's not as disruptive as Jmicron's case
and as long as ahci ignores class code, it doesn't really matter.
Driver can be chosen by changing loading order - this is both plus and
minus.
* As Arjan pointed out, that unlock-modify-lock sequence should be done
on resume too. ahci_init_controller() is the right place for such
stuff. This chip is going into notebooks, right?
--
tejun
> * Unlike Jmicron's case, this doesn't affect PCI bus scan. Actually, it
> does change class code but that's not as disruptive as Jmicron's case
> and as long as ahci ignores class code, it doesn't really matter.
> Driver can be chosen by changing loading order - this is both plus and
> minus.
The load order is basically undefined. You want AHCI so we should do
this early. That means either putting the same gunk all over the kernel
(drivers/ide, drivers/ata/*ati* drivers/ata/ahci) or in one place.
> * As Arjan pointed out, that unlock-modify-lock sequence should be done
> on resume too.
The infrastructure for this is already handled by the pci resume quirk
patches I sent.
Alan wrote:
>> * Unlike Jmicron's case, this doesn't affect PCI bus scan. Actually, it
>> does change class code but that's not as disruptive as Jmicron's case
>> and as long as ahci ignores class code, it doesn't really matter.
>> Driver can be chosen by changing loading order - this is both plus and
>> minus.
>
> The load order is basically undefined. You want AHCI so we should do
> this early. That means either putting the same gunk all over the kernel
> (drivers/ide, drivers/ata/*ati* drivers/ata/ahci) or in one place.
>
>> * As Arjan pointed out, that unlock-modify-lock sequence should be done
>> on resume too.
>
> The infrastructure for this is already handled by the pci resume quirk
> patches I sent.
As long as resume is properly handled, no problem.
Thanks.
--
tejun