[U-Boot] [PATCH] 74xx: use r4 instead of r2 in lock_ram_in_cache and unlock_ram_in_cache
The patch is following the commit 392438406041415fe64ab8748ec5ab5ad01d1cf7
mpc86xx: use r4 instead of r2 in lock_ram_in_cache and unlock_ram_in_cache
This is needed in unlock_ram_in_cache() because it is called from C and will corrupt the small data area anchor that is kept in R2.
lock_ram_in_cache() is modified similarly as good coding practice, but is not called from C.
Signed-off-by: Nick Spence nick.spence@freescale.com
also, the r2 is used as global data pointer.
Signed-off-by: Dave Liu daveliu@freescale.com --- cpu/74xx_7xx/start.S | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/cpu/74xx_7xx/start.S b/cpu/74xx_7xx/start.S index 07bbe01..b5484e3 100644 --- a/cpu/74xx_7xx/start.S +++ b/cpu/74xx_7xx/start.S @@ -857,9 +857,9 @@ lock_ram_in_cache: */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l - li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ + li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32 - mtctr r2 + mtctr r4 1: dcbz r0, r3 addi r3, r3, 32 @@ -878,9 +878,9 @@ unlock_ram_in_cache: /* invalidate the INIT_RAM section */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l - li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ + li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32 - mtctr r2 + mtctr r4 1: icbi r0, r3 addi r3, r3, 32 bdnz 1b
On Oct 23, 2008, at 8:59 AM, Dave Liu wrote:
The patch is following the commit 392438406041415fe64ab8748ec5ab5ad01d1cf7
mpc86xx: use r4 instead of r2 in lock_ram_in_cache and unlock_ram_in_cache
This is needed in unlock_ram_in_cache() because it is called from C and will corrupt the small data area anchor that is kept in R2.
lock_ram_in_cache() is modified similarly as good coding practice, but is not called from C.
Signed-off-by: Nick Spence nick.spence@freescale.com
also, the r2 is used as global data pointer.
Signed-off-by: Dave Liu daveliu@freescale.com
cpu/74xx_7xx/start.S | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/cpu/74xx_7xx/start.S b/cpu/74xx_7xx/start.S index 07bbe01..b5484e3 100644 --- a/cpu/74xx_7xx/start.S +++ b/cpu/74xx_7xx/start.S @@ -857,9 +857,9 @@ lock_ram_in_cache: */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
- li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \
- li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
- mtctr r2
- mtctr r4
1: dcbz r0, r3 addi r3, r3, 32 @@ -878,9 +878,9 @@ unlock_ram_in_cache: /* invalidate the INIT_RAM section */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
- li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \
- li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
- mtctr r2
- mtctr r4
1: icbi r0, r3 addi r3, r3, 32 bdnz 1b
Can we change the 31 to CONFIG_SYS_CACHELINE_SIZE-1
It doesn't matter much for 7xx/74xx/e600 as the cache line size has always been 32-bytes, but good to make the code a bit more readable
- k
From: Kumar Gala [mailto:galak@kernel.crashing.org] On Oct 23, 2008, at 8:59 AM, Dave Liu wrote:
The patch is following the commit 392438406041415fe64ab8748ec5ab5ad01d1cf7
mpc86xx: use r4 instead of r2 in lock_ram_in_cache and unlock_ram_in_cache
This is needed in unlock_ram_in_cache() because it is
called from C
and will corrupt the small data area anchor that is kept in R2.
lock_ram_in_cache() is modified similarly as good coding
practice, but
is not called from C.
Signed-off-by: Nick Spence nick.spence@freescale.com
also, the r2 is used as global data pointer.
Signed-off-by: Dave Liu daveliu@freescale.com
cpu/74xx_7xx/start.S | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/cpu/74xx_7xx/start.S b/cpu/74xx_7xx/start.S index 07bbe01..b5484e3 100644 --- a/cpu/74xx_7xx/start.S +++ b/cpu/74xx_7xx/start.S @@ -857,9 +857,9 @@ lock_ram_in_cache: */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
- li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \
- li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
- mtctr r2
- mtctr r4
1: dcbz r0, r3 addi r3, r3, 32 @@ -878,9 +878,9 @@ unlock_ram_in_cache: /* invalidate the INIT_RAM section */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
- li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \
- li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
- mtctr r2
- mtctr r4
1: icbi r0, r3 addi r3, r3, 32 bdnz 1b
Can we change the 31 to CONFIG_SYS_CACHELINE_SIZE-1
It doesn't matter much for 7xx/74xx/e600 as the cache line size has always been 32-bytes, but good to make the code a bit more readable
Patch welcomed!
Thanks, Dave
On Oct 23, 2008, at 9:20 AM, Liu Dave-R63238 wrote:
From: Kumar Gala [mailto:galak@kernel.crashing.org] On Oct 23, 2008, at 8:59 AM, Dave Liu wrote:
The patch is following the commit 392438406041415fe64ab8748ec5ab5ad01d1cf7
mpc86xx: use r4 instead of r2 in lock_ram_in_cache and unlock_ram_in_cache
This is needed in unlock_ram_in_cache() because it is
called from C
and will corrupt the small data area anchor that is kept in R2.
lock_ram_in_cache() is modified similarly as good coding
practice, but
is not called from C.
Signed-off-by: Nick Spence nick.spence@freescale.com
also, the r2 is used as global data pointer.
Signed-off-by: Dave Liu daveliu@freescale.com
cpu/74xx_7xx/start.S | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/cpu/74xx_7xx/start.S b/cpu/74xx_7xx/start.S index 07bbe01..b5484e3 100644 --- a/cpu/74xx_7xx/start.S +++ b/cpu/74xx_7xx/start.S @@ -857,9 +857,9 @@ lock_ram_in_cache: */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
- li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \
- li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
- mtctr r2
- mtctr r4
1: dcbz r0, r3 addi r3, r3, 32 @@ -878,9 +878,9 @@ unlock_ram_in_cache: /* invalidate the INIT_RAM section */ lis r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@h ori r3, r3, (CONFIG_SYS_INIT_RAM_ADDR & ~31)@l
- li r2, ((CONFIG_SYS_INIT_RAM_END & ~31) + \
- li r4, ((CONFIG_SYS_INIT_RAM_END & ~31) + \ (CONFIG_SYS_INIT_RAM_ADDR & 31) + 31) / 32
- mtctr r2
- mtctr r4
1: icbi r0, r3 addi r3, r3, 32 bdnz 1b
Can we change the 31 to CONFIG_SYS_CACHELINE_SIZE-1
It doesn't matter much for 7xx/74xx/e600 as the cache line size has always been 32-bytes, but good to make the code a bit more readable
Patch welcomed!
I'm willing to deal with it that way.
Ack on this patch.
- k
Dear Dave Liu,
In message 1224770375-30288-1-git-send-email-daveliu@freescale.com you wrote:
The patch is following the commit 392438406041415fe64ab8748ec5ab5ad01d1cf7
mpc86xx: use r4 instead of r2 in lock_ram_in_cache and unlock_ram_in_cache
This is needed in unlock_ram_in_cache() because it is called from C and will corrupt the small data area anchor that is kept in R2.
lock_ram_in_cache() is modified similarly as good coding practice, but is not called from C.
Signed-off-by: Nick Spence nick.spence@freescale.com
also, the r2 is used as global data pointer.
Signed-off-by: Dave Liu daveliu@freescale.com
cpu/74xx_7xx/start.S | 8 ++++---- 1 files changed, 4 insertions(+), 4 deletions(-)
Applied, thanks.
Best regards,
Wolfgang Denk
participants (4)
-
Dave Liu -
Kumar Gala -
Liu Dave-R63238
-
Wolfgang Denk