* [PATCH] TX49 has write buffer
@ 2006-08-18 23:00 Sergei Shtylyov
2006-08-19 14:11 ` Atsushi Nemoto
2006-08-19 16:36 ` Sergei Shtylyov
0 siblings, 2 replies; 5+ messages in thread
From: Sergei Shtylyov @ 2006-08-18 23:00 UTC (permalink / raw)
To: Linux-MIPS, Ralf Baechle; +Cc: Manish Lachwani
[-- Attachment #1: Type: text/plain, Size: 180 bytes --]
TX49 CPUs have a write buffer, so we need to select CPU_HAS_WB -- otherwise
all Toshiba RBTX49xx kernels fail to build.
Signed-off-by: Sergei Shtylyov <sshtylyov@ru.mvista.com>
[-- Attachment #2: TX49-has-write-buffer.patch --]
[-- Type: text/plain, Size: 401 bytes --]
Index: linux-mips/arch/mips/Kconfig
===================================================================
--- linux-mips.orig/arch/mips/Kconfig
+++ linux-mips/arch/mips/Kconfig
@@ -1225,6 +1225,7 @@ config CPU_TX49XX
bool "R49XX"
depends on SYS_HAS_CPU_TX49XX
select CPU_HAS_LLSC
+ select CPU_HAS_WB
select CPU_HAS_PREFETCH
select CPU_SUPPORTS_32BIT_KERNEL
select CPU_SUPPORTS_64BIT_KERNEL
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] TX49 has write buffer
2006-08-18 23:00 [PATCH] TX49 has write buffer Sergei Shtylyov
@ 2006-08-19 14:11 ` Atsushi Nemoto
2006-08-19 16:04 ` Sergei Shtylyov
2006-08-19 16:36 ` Sergei Shtylyov
1 sibling, 1 reply; 5+ messages in thread
From: Atsushi Nemoto @ 2006-08-19 14:11 UTC (permalink / raw)
To: sshtylyov; +Cc: linux-mips, ralf, mlachwani
On Sat, 19 Aug 2006 03:00:23 +0400, Sergei Shtylyov <sshtylyov@ru.mvista.com> wrote:
> TX49 CPUs have a write buffer, so we need to select CPU_HAS_WB -- otherwise
> all Toshiba RBTX49xx kernels fail to build.
TX49 CPUs also have a SYNC instruction which flushes a write buffer.
I think it is enough and wbflush() have been abused in
arch/mips/tx4927/ and arch/mips/tx4938/ codes.
There is old thread about this issue:
http://www.linux-mips.org/cgi-bin/mesg.cgi?a=linux-mips&i=Pine.GSO.3.96.1030415161611.13254H-100000%40delta.ds2.pg.gda.pl
---
Atsushi Nemoto
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] TX49 has write buffer
2006-08-19 14:11 ` Atsushi Nemoto
@ 2006-08-19 16:04 ` Sergei Shtylyov
0 siblings, 0 replies; 5+ messages in thread
From: Sergei Shtylyov @ 2006-08-19 16:04 UTC (permalink / raw)
To: Atsushi Nemoto; +Cc: linux-mips, ralf, mlachwani
Hello.
Atsushi Nemoto wrote:
>>TX49 CPUs have a write buffer, so we need to select CPU_HAS_WB -- otherwise
>>all Toshiba RBTX49xx kernels fail to build.
> TX49 CPUs also have a SYNC instruction which flushes a write buffer.
> I think it is enough and wbflush() have been abused in
> arch/mips/tx4927/ and arch/mips/tx4938/ codes.
How about a patch? I needed the kernel up and running, so I came up with
the obvious patch. I don't have no time to fix this assumed abuse.
I should also note, that this patch wasn't enough to bring RBTX4938 kernel
back to life since rbhma4500_defconfig is broken somewhere so the kernel
doesn't output anything on the colsole). If I have some more time, I'll try to
investigate what exactly was causing this (I have a working .config)...
WBR, Sergei
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] TX49 has write buffer
2006-08-18 23:00 [PATCH] TX49 has write buffer Sergei Shtylyov
2006-08-19 14:11 ` Atsushi Nemoto
@ 2006-08-19 16:36 ` Sergei Shtylyov
2006-08-20 13:55 ` Atsushi Nemoto
1 sibling, 1 reply; 5+ messages in thread
From: Sergei Shtylyov @ 2006-08-19 16:36 UTC (permalink / raw)
To: Linux-MIPS; +Cc: Ralf Baechle, Manish Lachwani, anemo
Hello.
Sergei Shtylyov wrote:
> TX49 CPUs have a write buffer, so we need to select CPU_HAS_WB --
> otherwise all Toshiba RBTX49xx kernels fail to build.
The uptimate reason is I think that <asm/system.h> doesn't include
<asm/wbflush.h> if CONFIG_CPU_HAS_WB is undefined -- although, <asm/wbflush.h>
handles this situation itself. Well, <asm/system.h> doesn't need the wbflush()
macro in that case, so it's in his own right to not include that header...
WBR, Sergei
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH] TX49 has write buffer
2006-08-19 16:36 ` Sergei Shtylyov
@ 2006-08-20 13:55 ` Atsushi Nemoto
0 siblings, 0 replies; 5+ messages in thread
From: Atsushi Nemoto @ 2006-08-20 13:55 UTC (permalink / raw)
To: sshtylyov; +Cc: linux-mips, ralf, mlachwani
On Sat, 19 Aug 2006 20:36:13 +0400, Sergei Shtylyov <sshtylyov@ru.mvista.com> wrote:
> The uptimate reason is I think that <asm/system.h> doesn't
> include <asm/wbflush.h> if CONFIG_CPU_HAS_WB is undefined --
> although, <asm/wbflush.h> handles this situation itself. Well,
> <asm/system.h> doesn't need the wbflush() macro in that case, so
> it's in his own right to not include that header...
Yes, so I think we should include <asm/wbflush.h> if we used
wbflush(). Here is my proposal patch.
Subject: Fix build errors related to wbflush.h on tx4927/tx4938.
TX49 CPUs have a SYNC instruction so that CONFIG_CPU_HAS_WB is no
longer needed.
Signed-off-by: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
diff --git a/arch/mips/tx4927/common/tx4927_setup.c b/arch/mips/tx4927/common/tx4927_setup.c
index 3ace403..4658b2a 100644
--- a/arch/mips/tx4927/common/tx4927_setup.c
+++ b/arch/mips/tx4927/common/tx4927_setup.c
@@ -53,19 +53,9 @@ void __init tx4927_time_init(void);
void dump_cp0(char *key);
-void (*__wbflush) (void);
-
-static void tx4927_write_buffer_flush(void)
-{
- __asm__ __volatile__
- ("sync\n\t" "nop\n\t" "loop: bc0f loop\n\t" "nop\n\t");
-}
-
-
void __init plat_mem_setup(void)
{
board_time_init = tx4927_time_init;
- __wbflush = tx4927_write_buffer_flush;
#ifdef CONFIG_TOSHIBA_RBTX4927
{
diff --git a/arch/mips/tx4927/toshiba_rbtx4927/toshiba_rbtx4927_irq.c b/arch/mips/tx4927/toshiba_rbtx4927/toshiba_rbtx4927_irq.c
index b0f021f..1b7cde1 100644
--- a/arch/mips/tx4927/toshiba_rbtx4927/toshiba_rbtx4927_irq.c
+++ b/arch/mips/tx4927/toshiba_rbtx4927/toshiba_rbtx4927_irq.c
@@ -130,6 +130,7 @@ #include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/reboot.h>
#include <asm/time.h>
+#include <asm/wbflush.h>
#include <linux/bootmem.h>
#include <linux/blkdev.h>
#ifdef CONFIG_RTC_DS1742
diff --git a/arch/mips/tx4938/common/irq.c b/arch/mips/tx4938/common/irq.c
index dc30d66..f2be0ea 100644
--- a/arch/mips/tx4938/common/irq.c
+++ b/arch/mips/tx4938/common/irq.c
@@ -30,6 +30,7 @@ #include <asm/io.h>
#include <asm/irq.h>
#include <asm/mipsregs.h>
#include <asm/system.h>
+#include <asm/wbflush.h>
#include <asm/tx4938/rbtx4938.h>
/**********************************************************************************/
diff --git a/arch/mips/tx4938/common/setup.c b/arch/mips/tx4938/common/setup.c
index 71859c4..f415a1f 100644
--- a/arch/mips/tx4938/common/setup.c
+++ b/arch/mips/tx4938/common/setup.c
@@ -41,29 +41,10 @@ void __init tx4938_setup(void);
void __init tx4938_time_init(void);
void dump_cp0(char *key);
-void (*__wbflush) (void);
-
-static void
-tx4938_write_buffer_flush(void)
-{
- mmiowb();
-
- __asm__ __volatile__(
- ".set push\n\t"
- ".set noreorder\n\t"
- "lw $0,%0\n\t"
- "nop\n\t"
- ".set pop"
- : /* no output */
- : "m" (*(int *)KSEG1)
- : "memory");
-}
-
void __init
plat_mem_setup(void)
{
board_time_init = tx4938_time_init;
- __wbflush = tx4938_write_buffer_flush;
toshiba_rbtx4938_setup();
}
diff --git a/arch/mips/tx4938/toshiba_rbtx4938/irq.c b/arch/mips/tx4938/toshiba_rbtx4938/irq.c
index 83f2750..8ef8c4e 100644
--- a/arch/mips/tx4938/toshiba_rbtx4938/irq.c
+++ b/arch/mips/tx4938/toshiba_rbtx4938/irq.c
@@ -84,6 +84,7 @@ #include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/reboot.h>
#include <asm/time.h>
+#include <asm/wbflush.h>
#include <linux/bootmem.h>
#include <asm/tx4938/rbtx4938.h>
^ permalink raw reply related [flat|nested] 5+ messages in thread
end of thread, other threads:[~2006-08-20 13:54 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2006-08-18 23:00 [PATCH] TX49 has write buffer Sergei Shtylyov
2006-08-19 14:11 ` Atsushi Nemoto
2006-08-19 16:04 ` Sergei Shtylyov
2006-08-19 16:36 ` Sergei Shtylyov
2006-08-20 13:55 ` Atsushi Nemoto
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.