[U-Boot] [PATCH] pci: Divided pci code of the powerpc
A source code of pci is written mainly on powerpc. Macro uses a pci source code of powerpc because other architecture (ex. SuperH) doesn't work.
I made asm-ppc/pci.h and moved macro of pci. Other archtecture need to make the following macro in asm-xxx/pci.h #define pci_phys_to_mem(dev, addr) (addr) #define pci_mem_to_phys(dev, addr) (addr) #define pci_phys_to_io(dev, addr) (addr) #define pci_io_to_phys(dev, addr) (addr)
Or have to make a function for pci address conversion. However, I want comment because I do not know the other architecture.
Signed-off-by: Nobuhiro Iwamatsu iwamatsu@nigauri.org --- include/asm-ppc/pci.h | 53 +++++++++++++++++++++++++++++++++++++++++++++++++ include/asm-sh/pci.h | 5 ++++ include/pci.h | 22 +------------------- 3 files changed, 59 insertions(+), 21 deletions(-) create mode 100644 include/asm-ppc/pci.h
diff --git a/include/asm-ppc/pci.h b/include/asm-ppc/pci.h new file mode 100644 index 0000000..d5f8475 --- /dev/null +++ b/include/asm-ppc/pci.h @@ -0,0 +1,53 @@ +/* + * (C) Copyright 2001 Sysgo Real-Time Solutions, GmbH <www.elinos.com> + * Andreas Heppel aheppel@sysgo.de + * + * (C) Copyright 2002 + * Wolfgang Denk, DENX Software Engineering, wd@denx.de. + * + * See file CREDITS for list of people who contributed to this + * project. + * + * 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 the Free Software Foundation; either version 2 of + * the License, or (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * aloong with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, + * MA 02111-1307 USA + */ + +#ifndef _ASM_PCI_H +#define _ASM_PCI_H + +extern phys_addr_t pci_hose_bus_to_phys(struct pci_controller* hose, + unsigned long addr, unsigned long flags); +extern unsigned long pci_hose_phys_to_bus(struct pci_controller* hose, + phys_addr_t addr, unsigned long flags); + +#define pci_phys_to_bus(dev, addr, flags) \ + pci_hose_phys_to_bus(pci_bus_to_hose(PCI_BUS(dev)), (addr), (flags)) +#define pci_bus_to_phys(dev, addr, flags) \ + pci_hose_bus_to_phys(pci_bus_to_hose(PCI_BUS(dev)), (addr), (flags)) + +#define pci_phys_to_mem(dev, addr) pci_phys_to_bus((dev), (addr), PCI_REGION_MEM) +#define pci_mem_to_phys(dev, addr) pci_bus_to_phys((dev), (addr), PCI_REGION_MEM) +#define pci_phys_to_io(dev, addr) pci_phys_to_bus((dev), (addr), PCI_REGION_IO) +#define pci_io_to_phys(dev, addr) pci_bus_to_phys((dev), (addr), PCI_REGION_IO) + +#ifdef CONFIG_MPC824X +extern void pci_mpc824x_init (struct pci_controller *hose); +#endif + +#ifdef CONFIG_MPC85xx +extern void pci_mpc85xx_init (struct pci_controller *hose); +#endif + +#endif /* _ASM_PCI_H */ diff --git a/include/asm-sh/pci.h b/include/asm-sh/pci.h index bc59491..1f24fb6 100644 --- a/include/asm-sh/pci.h +++ b/include/asm-sh/pci.h @@ -44,4 +44,9 @@ int pci_sh4_read_config_dword(struct pci_controller *hose, int pci_sh4_write_config_dword(struct pci_controller *hose, pci_dev_t dev, int offset, u32 value);
+#define pci_phys_to_mem(dev, addr) (addr) +#define pci_mem_to_phys(dev, addr) (addr) +#define pci_phys_to_io(dev, addr) (addr) +#define pci_io_to_phys(dev, addr) (addr) + #endif /* _ASM_PCI_H_ */ diff --git a/include/pci.h b/include/pci.h index 1c8e216..9b61635 100644 --- a/include/pci.h +++ b/include/pci.h @@ -432,21 +432,6 @@ extern __inline__ void pci_set_ops(struct pci_controller *hose,
extern void pci_setup_indirect(struct pci_controller* hose, u32 cfg_addr, u32 cfg_data);
-extern phys_addr_t pci_hose_bus_to_phys(struct pci_controller* hose, - unsigned long addr, unsigned long flags); -extern unsigned long pci_hose_phys_to_bus(struct pci_controller* hose, - phys_addr_t addr, unsigned long flags); - -#define pci_phys_to_bus(dev, addr, flags) \ - pci_hose_phys_to_bus(pci_bus_to_hose(PCI_BUS(dev)), (addr), (flags)) -#define pci_bus_to_phys(dev, addr, flags) \ - pci_hose_bus_to_phys(pci_bus_to_hose(PCI_BUS(dev)), (addr), (flags)) - -#define pci_phys_to_mem(dev, addr) pci_phys_to_bus((dev), (addr), PCI_REGION_MEM) -#define pci_mem_to_phys(dev, addr) pci_bus_to_phys((dev), (addr), PCI_REGION_MEM) -#define pci_phys_to_io(dev, addr) pci_phys_to_bus((dev), (addr), PCI_REGION_IO) -#define pci_io_to_phys(dev, addr) pci_bus_to_phys((dev), (addr), PCI_REGION_IO) - extern int pci_hose_read_config_byte(struct pci_controller *hose, pci_dev_t dev, int where, u8 *val); extern int pci_hose_read_config_word(struct pci_controller *hose, @@ -503,11 +488,6 @@ extern int pci_hose_config_device(struct pci_controller *hose, unsigned long mem, unsigned long command);
-#ifdef CONFIG_MPC824X -extern void pci_mpc824x_init (struct pci_controller *hose); -#endif +#include <asm/pci.h>
-#ifdef CONFIG_MPC85xx -extern void pci_mpc85xx_init (struct pci_controller *hose); -#endif #endif /* _PCI_H */
Dear Nobuhiro Iwamatsu,
In message 20080829103949.022a6e45.iwamatsu@nigauri.org you wrote:
A source code of pci is written mainly on powerpc. Macro uses a pci source code of powerpc because other architecture (ex. SuperH) doesn't work.
I made asm-ppc/pci.h and moved macro of pci. Other archtecture need to make the following macro in asm-xxx/pci.h #define pci_phys_to_mem(dev, addr) (addr) #define pci_mem_to_phys(dev, addr) (addr) #define pci_phys_to_io(dev, addr) (addr) #define pci_io_to_phys(dev, addr) (addr)
Or have to make a function for pci address conversion. However, I want comment because I do not know the other architecture.
To me the patch looks OK, but I'm not an expert for al the affected architectures.
Hello everybody, please comment!!
Otherwise I will just apply the patch and then not listen to your complaints if it breaks anything.
Best regards,
Wolfgang Denk
Hi, Wolfgang .
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Best regards, Nobuhiro
2008/9/7 Wolfgang Denk wd@denx.de:
Dear Nobuhiro Iwamatsu,
In message 20080829103949.022a6e45.iwamatsu@nigauri.org you wrote:
A source code of pci is written mainly on powerpc. Macro uses a pci source code of powerpc because other architecture (ex. SuperH) doesn't work.
I made asm-ppc/pci.h and moved macro of pci. Other archtecture need to make the following macro in asm-xxx/pci.h #define pci_phys_to_mem(dev, addr) (addr) #define pci_mem_to_phys(dev, addr) (addr) #define pci_phys_to_io(dev, addr) (addr) #define pci_io_to_phys(dev, addr) (addr)
Or have to make a function for pci address conversion. However, I want comment because I do not know the other architecture.
To me the patch looks OK, but I'm not an expert for al the affected architectures.
Hello everybody, please comment!!
Otherwise I will just apply the patch and then not listen to your complaints if it breaks anything.
Best regards,
Wolfgang Denk
-- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de History tends to exaggerate. -- Col. Green, "The Savage Curtain", stardate 5906.4 _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
Dear Wolfgang,
I'm doing some work on i386 at the moment and: a) It seems PCI is already broken - see [PATCH 0/2] Update i386 code (sc520_cdp) (http://lists.denx.de/pipermail/u-boot/2008-September/040099.html) b) The board I'm porting to doesn't use PCI c) I don't have a sc520 board with PCI to test on
My vote would be to commit and fix up i386 later when someone needs PCI
Harsh, but true
Regards,
Graeme
On Thu, Sep 18, 2008 at 1:36 PM, Nobuhiro Iwamatsu iwamatsu@nigauri.org wrote:
Hi, Wolfgang .
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Best regards, Nobuhiro
2008/9/7 Wolfgang Denk wd@denx.de:
Dear Nobuhiro Iwamatsu,
In message 20080829103949.022a6e45.iwamatsu@nigauri.org you wrote:
A source code of pci is written mainly on powerpc. Macro uses a pci source code of powerpc because other architecture (ex. SuperH) doesn't work.
I made asm-ppc/pci.h and moved macro of pci. Other archtecture need to make the following macro in asm-xxx/pci.h #define pci_phys_to_mem(dev, addr) (addr) #define pci_mem_to_phys(dev, addr) (addr) #define pci_phys_to_io(dev, addr) (addr) #define pci_io_to_phys(dev, addr) (addr)
Or have to make a function for pci address conversion. However, I want comment because I do not know the other architecture.
To me the patch looks OK, but I'm not an expert for al the affected architectures.
Hello everybody, please comment!!
Otherwise I will just apply the patch and then not listen to your complaints if it breaks anything.
Best regards,
Wolfgang Denk
-- DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: (+49)-8142-66989-10 Fax: (+49)-8142-66989-80 Email: wd@denx.de History tends to exaggerate. -- Col. Green, "The Savage Curtain", stardate 5906.4 _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
-- Nobuhiro Iwamatsu _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
Dear Stefan, Kim, Andy, Jon & Kumar,
in message bc6f41e00809172036o7b4c7408l70a7960f9f3995f1@mail.gmail.com Nobuhiro Iwamatsu wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
Best regards,
Wolfgang Denk
On Mon, Sep 22, 2008 at 3:11 PM, Wolfgang Denk wd@denx.de wrote:
Dear Stefan, Kim, Andy, Jon & Kumar,
in message bc6f41e00809172036o7b4c7408l70a7960f9f3995f1@mail.gmail.com Nobuhiro Iwamatsu wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
(forgot to reply all on this, already sent to the O.P.)
I think this is going to break some boards/archs. From the top of the u-boot tree, if you do
find . -type f -exec grep pci_register_hose {} /dev/null ;
it shows all the source files that are registering a PCI bus. A quick look shows there are MIPS, ARM, x86 in that list at least, that aren't addressed by this change.
On Mon, Sep 22, 2008 at 3:33 PM, Andrew Dyer amdyer@gmail.com wrote:
On Mon, Sep 22, 2008 at 3:11 PM, Wolfgang Denk wd@denx.de wrote:
Dear Stefan, Kim, Andy, Jon & Kumar,
in message bc6f41e00809172036o7b4c7408l70a7960f9f3995f1@mail.gmail.com Nobuhiro Iwamatsu wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
If some of that code is powerpc-only, maybe we should move the code out of pci.c, too? pci_hose_phys_to_bus() appears to be... board/eltec/bab7xx/pci.c only, actually. The declaration was moved to asm/pci.h, but the implementation is in drivers/pci/pci.c.
Andy
On Tuesday 23 September 2008, Andy Fleming wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
If some of that code is powerpc-only, maybe we should move the code out of pci.c, too? pci_hose_phys_to_bus() appears to be... board/eltec/bab7xx/pci.c only, actually.
That doesn't seem to be correct. In include/pci.h:
#define pci_phys_to_bus(dev, addr, flags) \ pci_hose_phys_to_bus(pci_bus_to_hose(PCI_BUS(dev)), (addr), (flags))
and later:
#define pci_phys_to_mem(dev, addr) pci_phys_to_bus((dev), (addr), PCI_REGION_MEM) #define pci_phys_to_io(dev, addr) pci_phys_to_bus((dev), (addr), PCI_REGION_IO)
So calling pci_phys_to_mem() or pci_phys_to_io() leads to a pci_hose_phys_to_bus() call in the end.
Best regards, Stefan
===================================================================== DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office@denx.de =====================================================================
On Monday 22 September 2008, Andrew Dyer wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
(forgot to reply all on this, already sent to the O.P.)
I think this is going to break some boards/archs. From the top of the u-boot tree, if you do
find . -type f -exec grep pci_register_hose {} /dev/null ;
it shows all the source files that are registering a PCI bus. A quick look shows there are MIPS, ARM, x86 in that list at least, that aren't addressed by this change.
That's correct. This patch moves the original defines to the PPC specific header and introduces new defines in the SH specific header. All other ARCH's will not compile anymore when they try to include PCI support.
So, NAK from me. An acceptable patch would be to move the original defines at least to all ARCH's already using PCI.
BTW: I fail to see what's really PPC specific about the pci code in question. Nobuhiro, could you please elaborate what's the problem here?
Thanks.
Best regards, Stefan
===================================================================== DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office@denx.de =====================================================================
Hi, all.
2008/9/23 Stefan Roese sr@denx.de:
On Monday 22 September 2008, Andrew Dyer wrote:
I did a simple check.
Only powerpc and i386 and sh seem to use PCI. If powerpc does not have a problem, would you commit it?
Do I have your "go" (or at least no veto) to apply this patch?
(forgot to reply all on this, already sent to the O.P.)
I think this is going to break some boards/archs. From the top of the u-boot tree, if you do
find . -type f -exec grep pci_register_hose {} /dev/null ;
it shows all the source files that are registering a PCI bus. A quick look shows there are MIPS, ARM, x86 in that list at least, that aren't addressed by this change.
That's correct. This patch moves the original defines to the PPC specific header and introduces new defines in the SH specific header. All other ARCH's will not compile anymore when they try to include PCI support.
I understand that some macro of the PCI does not depend on only PPC as Andrew points it out. I am thinking other countermeasure.
So, NAK from me. An acceptable patch would be to move the original defines at least to all ARCH's already using PCI.
BTW: I fail to see what's really PPC specific about the pci code in question. Nobuhiro, could you please elaborate what's the problem here?
I wanted to remove CPU and baords of specific code from from PCI network driver. CONFIG_E500, DB64360 and DB64460 CPU is PPC. I thought that I could collect these with header files of PPC. Therefore I sent that patch in RFC.
Best regards, Nobuhiro
On Wednesday 24 September 2008, Nobuhiro Iwamatsu wrote:
BTW: I fail to see what's really PPC specific about the pci code in question. Nobuhiro, could you please elaborate what's the problem here?
I wanted to remove CPU and baords of specific code from from PCI network driver. CONFIG_E500, DB64360 and DB64460 CPU is PPC.
Now I understand. You are referring to code like this:
drivers/net/eepro100.c:
#if defined(CONFIG_E500) || defined(CONFIG_DB64360) || defined(CONFIG_DB64460) #define bus_to_phys(a) (a) #define phys_to_bus(a) (a) #else #define bus_to_phys(a) pci_mem_to_phys((pci_dev_t)dev->priv, a) #define phys_to_bus(a) pci_phys_to_mem((pci_dev_t)dev->priv, a) #endif
I thought that I could collect these with header files of PPC. Therefore I sent that patch in RFC.
Yes, it definitely is a good idea to move those defines to platform/cpu headers.
Best regards, Stefan
===================================================================== DENX Software Engineering GmbH, MD: Wolfgang Denk & Detlev Zundel HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany Phone: +49-8142-66989-0 Fax: +49-8142-66989-80 Email: office@denx.de =====================================================================
participants (6)
-
Andrew Dyer -
Andy Fleming -
Graeme Russ -
Nobuhiro Iwamatsu -
Stefan Roese -
Wolfgang Denk