[patch] 2.4: Revert interface removal

[patch] 2.4: Revert interface removal

[Maciej W. Rozycki]

Hello,

 A recent change to include/asm-mips/scatterlist.h broke
drivers/scsi/dec_esp.c.  Since 2.4.19 is not the proper version to remove
interfaces, I'm going to check in the following patch to the 2.4 branch to
revert the change (with a slightly sanitized type for the dvma_address
member). 

 Any objections?

  Maciej

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +

patch-mips-2.4.19-rc1-20020802-sg-2
diff -up --recursive --new-file linux-mips-2.4.19-rc1-20020802.macro/include/asm-mips/scatterlist.h linux-mips-2.4.19-rc1-20020802/include/asm-mips/scatterlist.h
--- linux-mips-2.4.19-rc1-20020802.macro/include/asm-mips/scatterlist.h	2002-08-01 15:57:53.000000000 +0000
+++ linux-mips-2.4.19-rc1-20020802/include/asm-mips/scatterlist.h	2002-08-03 20:43:35.000000000 +0000
@@ -10,6 +10,13 @@ struct scatterlist {
 	unsigned int length;
 };
 
+struct mmu_sglist {
+	char *addr;
+	char *__dont_touch;
+	unsigned int len;
+	dma_addr_t dvma_address;
+};
+
 #define ISA_DMA_THRESHOLD (0x00ffffff)
 
 #endif /* __ASM_SCATTERLIST_H */
diff -up --recursive --new-file linux-mips-2.4.19-rc1-20020802.macro/include/asm-mips64/scatterlist.h linux-mips-2.4.19-rc1-20020802/include/asm-mips64/scatterlist.h
--- linux-mips-2.4.19-rc1-20020802.macro/include/asm-mips64/scatterlist.h	2002-08-01 15:57:54.000000000 +0000
+++ linux-mips-2.4.19-rc1-20020802/include/asm-mips64/scatterlist.h	2002-08-03 20:43:44.000000000 +0000
@@ -10,6 +10,13 @@ struct scatterlist {
 	unsigned long length;
 };
 
+struct mmu_sglist {
+	char *addr;
+	char *__dont_touch;
+	unsigned long len;
+	dma_addr_t dvma_address;
+};
+
 #define ISA_DMA_THRESHOLD (0x00ffffff)
 
 #endif /* __ASM_SCATTERLIST_H */

Re: [patch] 2.4: Revert interface removal

[Ralf Baechle]

On Mon, Aug 05, 2002 at 11:05:40AM +0200, Maciej W. Rozycki wrote:

>  A recent change to include/asm-mips/scatterlist.h broke
> drivers/scsi/dec_esp.c.  Since 2.4.19 is not the proper version to remove
> interfaces, I'm going to check in the following patch to the 2.4 branch to
> revert the change (with a slightly sanitized type for the dvma_address
> member). 
> 
>  Any objections?

Sorry for simply removing the structure, that was an accident.  The
question why the use of struct mmu_sglist still hasn't been replaced by
newer interfaces stays ...

So please go ahead and commit your patch.

  Ralf

Re: [patch] 2.4: Revert interface removal

[Carsten Langgaard]

Changing the scatterlist and the pci functions seems to break things in the IDE
interface.

/Carsten


Ralf Baechle wrote:

> On Mon, Aug 05, 2002 at 11:05:40AM +0200, Maciej W. Rozycki wrote:
>
> >  A recent change to include/asm-mips/scatterlist.h broke
> > drivers/scsi/dec_esp.c.  Since 2.4.19 is not the proper version to remove
> > interfaces, I'm going to check in the following patch to the 2.4 branch to
> > revert the change (with a slightly sanitized type for the dvma_address
> > member).
> >
> >  Any objections?
>
> Sorry for simply removing the structure, that was an accident.  The
> question why the use of struct mmu_sglist still hasn't been replaced by
> newer interfaces stays ...
>
> So please go ahead and commit your patch.
>
>   Ralf

--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com

Re: [patch] 2.4: Revert interface removal

[Carsten Langgaard]

Ralf, could we please revert the latest changes to include/asm-mips/scatterlist.h
and include/asm-mips/pci.h

/Carsten


Carsten Langgaard wrote:

> Changing the scatterlist and the pci functions seems to break things in the IDE
> interface.
>
> /Carsten
>
> Ralf Baechle wrote:
>
> > On Mon, Aug 05, 2002 at 11:05:40AM +0200, Maciej W. Rozycki wrote:
> >
> > >  A recent change to include/asm-mips/scatterlist.h broke
> > > drivers/scsi/dec_esp.c.  Since 2.4.19 is not the proper version to remove
> > > interfaces, I'm going to check in the following patch to the 2.4 branch to
> > > revert the change (with a slightly sanitized type for the dvma_address
> > > member).
> > >
> > >  Any objections?
> >
> > Sorry for simply removing the structure, that was an accident.  The
> > question why the use of struct mmu_sglist still hasn't been replaced by
> > newer interfaces stays ...
> >
> > So please go ahead and commit your patch.
> >
> >   Ralf
>
> --
> _    _ ____  ___   Carsten Langgaard  Mailto:carstenl@mips.com
> |\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
> | \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
>   TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
>                    Denmark            http://www.mips.com

--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com

Re: [patch] 2.4: Revert interface removal

[Carsten Langgaard]

[-- Attachment #1: Type: text/plain, Size: 2110 bytes --]

Ok, I finally figured out what the problem is.
The attached patch fix the problems, please apply.

/Carsten



Carsten Langgaard wrote:

> Ralf, could we please revert the latest changes to include/asm-mips/scatterlist.h
> and include/asm-mips/pci.h
>
> /Carsten
>
> Carsten Langgaard wrote:
>
> > Changing the scatterlist and the pci functions seems to break things in the IDE
> > interface.
> >
> > /Carsten
> >
> > Ralf Baechle wrote:
> >
> > > On Mon, Aug 05, 2002 at 11:05:40AM +0200, Maciej W. Rozycki wrote:
> > >
> > > >  A recent change to include/asm-mips/scatterlist.h broke
> > > > drivers/scsi/dec_esp.c.  Since 2.4.19 is not the proper version to remove
> > > > interfaces, I'm going to check in the following patch to the 2.4 branch to
> > > > revert the change (with a slightly sanitized type for the dvma_address
> > > > member).
> > > >
> > > >  Any objections?
> > >
> > > Sorry for simply removing the structure, that was an accident.  The
> > > question why the use of struct mmu_sglist still hasn't been replaced by
> > > newer interfaces stays ...
> > >
> > > So please go ahead and commit your patch.
> > >
> > >   Ralf
> >
> > --
> > _    _ ____  ___   Carsten Langgaard Mailto:carstenl@mips.com
> > |\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
> > | \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
> >   TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
> >                    Denmark           http://www.mips.com
>
> --
> _    _ ____  ___   Carsten Langgaard  Mailto:carstenl@mips.com
> |\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
> | \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
>   TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
>                    Denmark            http://www.mips.com

--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com



[-- Attachment #2: pci.3.patch --]
[-- Type: text/plain, Size: 2183 bytes --]

Index: arch/mips/kernel/pci-dma.c
===================================================================
RCS file: /cvs/linux/arch/mips/kernel/pci-dma.c,v
retrieving revision 1.7.2.1
diff -u -r1.7.2.1 pci-dma.c
--- arch/mips/kernel/pci-dma.c	2002/08/01 12:40:14	1.7.2.1
+++ arch/mips/kernel/pci-dma.c	2002/08/05 12:54:19
@@ -30,11 +30,11 @@
 
 	if (ret != NULL) {
 		memset(ret, 0, size);
+		*dma_handle = virt_to_bus(ret);
 #ifdef CONFIG_NONCOHERENT_IO
 		dma_cache_wback_inv((unsigned long) ret, size);
 		ret = UNCAC_ADDR(ret);
 #endif
-		*dma_handle = virt_to_bus(ret);
 	}
 
 	return ret;
Index: include/asm-mips/page.h
===================================================================
RCS file: /cvs/linux/include/asm-mips/page.h,v
retrieving revision 1.14.2.9
diff -u -r1.14.2.9 page.h
--- include/asm-mips/page.h	2002/08/01 12:40:14	1.14.2.9
+++ include/asm-mips/page.h	2002/08/05 12:54:32
@@ -125,8 +125,8 @@
 #define VM_DATA_DEFAULT_FLAGS  (VM_READ | VM_WRITE | VM_EXEC | \
 				VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
 
-#define UNCAC_ADDR(addr)	((addr) - (PAGE_OFFSET + UNCAC_BASE))
-#define CAC_ADDR(addr)		((addr) - (UNCAC_BASE + PAGE_OFFSET))
+#define UNCAC_ADDR(addr)	((addr) - PAGE_OFFSET + UNCAC_BASE)
+#define CAC_ADDR(addr)		((addr) - UNCAC_BASE + PAGE_OFFSET)
 
 /*
  * Memory above this physical address will be considered highmem.
Index: include/asm-mips/pci.h
===================================================================
RCS file: /cvs/linux/include/asm-mips/pci.h,v
retrieving revision 1.24.2.7
diff -u -r1.24.2.7 pci.h
--- include/asm-mips/pci.h	2002/08/01 12:40:14	1.24.2.7
+++ include/asm-mips/pci.h	2002/08/05 12:54:32
@@ -180,7 +180,7 @@
 		else if (!sg->address && !sg->page)
 			out_of_line_bug();
 
-		if (sg[i].address) {
+		if (sg->address) {
 			dma_cache_wback_inv((unsigned long)sg->address,
 			                    sg->length);
 			sg->dma_address = virt_to_bus(sg->address);
@@ -317,7 +317,7 @@
  * returns, or alternatively stop on the first sg_dma_len(sg) which
  * is 0.
  */
-#define sg_dma_address(sg)	((sg)->address)
+#define sg_dma_address(sg)	((sg)->dma_address)
 #define sg_dma_len(sg)		((sg)->length)
 
 #endif /* __KERNEL__ */

Re: [patch] 2.4: Revert interface removal

[Ralf Baechle]

On Mon, Aug 05, 2002 at 03:04:13PM +0200, Carsten Langgaard wrote:

> Ok, I finally figured out what the problem is.
> The attached patch fix the problems, please apply.

Applied, along with the 64-bit and 2.5 bits your patch was missing.

  Ralf

Re: [patch] 2.4: Revert interface removal

[Carsten Langgaard]

[-- Attachment #1: Type: text/plain, Size: 793 bytes --]

Ralf Baechle wrote:

> On Mon, Aug 05, 2002 at 03:04:13PM +0200, Carsten Langgaard wrote:
>
> > Ok, I finally figured out what the problem is.
> > The attached patch fix the problems, please apply.
>
> Applied, along with the 64-bit and 2.5 bits your patch was missing.
>

I was waiting for you to fix the bus_to_baddr in the 64-bit, I can see
you have have done something about it now, but I'm afraid you didn't get
it quite right.
Here is a patch to fix the typos.

/Carsten





--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com



[-- Attachment #2: pci.h.patch --]
[-- Type: text/plain, Size: 900 bytes --]

Index: include/asm-mips64/pci.h
===================================================================
RCS file: /cvs/linux/include/asm-mips64/pci.h,v
retrieving revision 1.16.2.9
diff -u -r1.16.2.9 pci.h
--- include/asm-mips64/pci.h	2002/08/06 02:21:50	1.16.2.9
+++ include/asm-mips64/pci.h	2002/08/06 06:36:35
@@ -213,8 +213,8 @@
 			out_of_line_bug();
 
 		dma_cache_wback_inv((unsigned long)sg->address, sg->length);
-		sg->address = bus_to_baddr(hwdev->bus->number) |
-		              virt_to_bus(sg->address);
+		sg->address = bus_to_baddr(hwdev->bus->number,
+					   virt_to_bus(sg->address));
 	}
 
 	return nents;
@@ -251,7 +251,7 @@
 	if (direction == PCI_DMA_NONE)
 		out_of_line_bug();
 
-	dma_cache_wback_inv((unsigned long)__va(dma_handle - bus_to_baddr(hwdev->bus->number)), size);
+	dma_cache_wback_inv((unsigned long)__va(bus_to_baddr(hwdev->bus->number, dma_handle)), size);
 }
 
 /*

Re: [patch] 2.4: Revert interface removal

[Carsten Langgaard]

[-- Attachment #1: Type: text/plain, Size: 2535 bytes --]

Carsten Langgaard wrote:

> Ralf Baechle wrote:
>
> > On Mon, Aug 05, 2002 at 03:04:13PM +0200, Carsten Langgaard wrote:
> >
> > > Ok, I finally figured out what the problem is.
> > > The attached patch fix the problems, please apply.
> >
> > Applied, along with the 64-bit and 2.5 bits your patch was missing.
> >
>

Another reason why I didn't send the patch against the 64-bit is that the pci handling is not completely
identical.
So the patch should actually look like this for things to work.



>
> I was waiting for you to fix the bus_to_baddr in the 64-bit, I can see
> you have have done something about it now, but I'm afraid you didn't get
> it quite right.
> Here is a patch to fix the typos.
>
> /Carsten
>
> --
> _    _ ____  ___   Carsten Langgaard  Mailto:carstenl@mips.com
> |\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
> | \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
>   TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
>                    Denmark            http://www.mips.com
>
>   ------------------------------------------------------------------------
> Index: include/asm-mips64/pci.h
> ===================================================================
> RCS file: /cvs/linux/include/asm-mips64/pci.h,v
> retrieving revision 1.16.2.9
> diff -u -r1.16.2.9 pci.h
> --- include/asm-mips64/pci.h    2002/08/06 02:21:50     1.16.2.9
> +++ include/asm-mips64/pci.h    2002/08/06 06:36:35
> @@ -213,8 +213,8 @@
>                         out_of_line_bug();
>
>                 dma_cache_wback_inv((unsigned long)sg->address, sg->length);
> -               sg->address = bus_to_baddr(hwdev->bus->number) |
> -                             virt_to_bus(sg->address);
> +               sg->address = bus_to_baddr(hwdev->bus->number,
> +                                          virt_to_bus(sg->address));
>         }
>
>         return nents;
> @@ -251,7 +251,7 @@
>         if (direction == PCI_DMA_NONE)
>                 out_of_line_bug();
>
> -       dma_cache_wback_inv((unsigned long)__va(dma_handle - bus_to_baddr(hwdev->bus->number)), size);
> +       dma_cache_wback_inv((unsigned long)__va(bus_to_baddr(hwdev->bus->number, dma_handle)), size);
>  }
>
>  /*

--
_    _ ____  ___   Carsten Langgaard   Mailto:carstenl@mips.com
|\  /|||___)(___   MIPS Denmark        Direct: +45 4486 5527
| \/ |||    ____)  Lautrupvang 4B      Switch: +45 4486 5555
  TECHNOLOGIES     2750 Ballerup       Fax...: +45 4486 5556
                   Denmark             http://www.mips.com



[-- Attachment #2: pci.h.patch --]
[-- Type: text/plain, Size: 1163 bytes --]

Index: include/asm-mips64/pci.h
===================================================================
RCS file: /cvs/linux/include/asm-mips64/pci.h,v
retrieving revision 1.16.2.9
diff -u -r1.16.2.9 pci.h
--- include/asm-mips64/pci.h	2002/08/06 02:21:50	1.16.2.9
+++ include/asm-mips64/pci.h	2002/08/06 09:01:25
@@ -213,8 +213,8 @@
 			out_of_line_bug();
 
 		dma_cache_wback_inv((unsigned long)sg->address, sg->length);
-		sg->address = bus_to_baddr(hwdev->bus->number) |
-		              virt_to_bus(sg->address);
+		sg->address = bus_to_baddr(hwdev->bus->number,
+					   virt_to_bus(sg->address));
 	}
 
 	return nents;
@@ -251,7 +251,7 @@
 	if (direction == PCI_DMA_NONE)
 		out_of_line_bug();
 
-	dma_cache_wback_inv((unsigned long)__va(dma_handle - bus_to_baddr(hwdev->bus->number)), size);
+	dma_cache_wback_inv((unsigned long)__va(bus_to_baddr(hwdev->bus->number, dma_handle)), size);
 }
 
 /*
@@ -341,7 +341,7 @@
  * returns, or alternatively stop on the first sg_dma_len(sg) which
  * is 0.
  */
-#define sg_dma_address(sg)	((sg)->dma_address)
+#define sg_dma_address(sg)	((sg)->address)
 #define sg_dma_len(sg)		((sg)->length)
 
 #endif /* __KERNEL__ */

Re: [patch] 2.4: Revert interface removal

[Maciej W. Rozycki]

On Mon, 5 Aug 2002, Ralf Baechle wrote:

> Sorry for simply removing the structure, that was an accident.  The
> question why the use of struct mmu_sglist still hasn't been replaced by
> newer interfaces stays ...

 Because the structure wasn't removed at the 2.3 time?  M68k still keeps
it as well and their sun3x_esp.c driver uses it (I think dec_esp.c was
derived from that one). 

 For 2.5, I'll work on adding an appropriate DMA addressing layer for the
TURBOchannel bus and the I/O ASIC and fixing the driver, of course. 

-- 
+  Maciej W. Rozycki, Technical University of Gdansk, Poland   +
+--------------------------------------------------------------+
+        e-mail: macro@ds2.pg.gda.pl, PGP key available        +