Patchwork [U-Boot,v2] arm: Tegra2: Fix ELDK42 gcc failure with inline asm stack pointer load

login
register
mail settings
Submitter Tom Warren
Date Feb. 17, 2012, 4:01 p.m.
Message ID <1329494481-8134-1-git-send-email-twarren@nvidia.com>
Download mbox | patch
Permalink /patch/141863/
State Accepted
Commit d8bd820935bb9b8bf2717a259eeab4376e9ccc9a
Headers show

Comments

Tom Warren - Feb. 17, 2012, 4:01 p.m.
The 4.2.2 gcc in the ELDK42 release doesn't like the direct SP
load using a constant in tegra2_start. Change it to use a load
thru another reg using mov sp, %0 : : "r"(CONST).

Tested on my Seaboard T20-A03, U-Boot loads and runs OK. Also
compiled all tegra2 builds with both gcc 4.2.2 and 4.4.1 OK.

Signed-off-by: Tom Warren <twarren@nvidia.com>
---
V2: Change 4 mov/orr's to single mov sp as per Marek Vasut

 arch/arm/cpu/armv7/tegra2/ap20.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)
Tom Warren - Feb. 23, 2012, 9:25 p.m.
On Fri, Feb 17, 2012 at 9:01 AM, Tom Warren <twarren.nvidia@gmail.com> wrote:
> The 4.2.2 gcc in the ELDK42 release doesn't like the direct SP
> load using a constant in tegra2_start. Change it to use a load
> thru another reg using mov sp, %0 : : "r"(CONST).
>
> Tested on my Seaboard T20-A03, U-Boot loads and runs OK. Also
> compiled all tegra2 builds with both gcc 4.2.2 and 4.4.1 OK.
>
> Signed-off-by: Tom Warren <twarren@nvidia.com>
> ---
> V2: Change 4 mov/orr's to single mov sp as per Marek Vasut
>
>  arch/arm/cpu/armv7/tegra2/ap20.c |   10 +++++-----
>  1 files changed, 5 insertions(+), 5 deletions(-)
>
> diff --git a/arch/arm/cpu/armv7/tegra2/ap20.c b/arch/arm/cpu/armv7/tegra2/ap20.c
> index 4c44bb3..f0450f4 100644
> --- a/arch/arm/cpu/armv7/tegra2/ap20.c
> +++ b/arch/arm/cpu/armv7/tegra2/ap20.c
> @@ -298,11 +298,11 @@ void tegra2_start(void)
>                writel(0xC0, &pmt->pmt_cfg_ctl);
>
>                /*
> -               * If we are ARM7 - give it a different stack. We are about to
> -               * start up the A9 which will want to use this one.
> -               */
> -               asm volatile("ldr       sp, =%c0\n"
> -                       : : "i"(AVP_EARLY_BOOT_STACK_LIMIT));
> +                * If we are ARM7 - give it a different stack. We are about to
> +                * start up the A9 which will want to use this one.
> +                */
> +               asm volatile("mov       sp, %0\n"
> +                       : : "r"(AVP_EARLY_BOOT_STACK_LIMIT));
>
>                start_cpu((u32)_start);
>                halt_avp();
> --
> 1.7.7.1
>

Ack? Nak? Can I get some love here?

Albert/Wolfgang - did anyone try this to ensure that it really fixes
the 6 Tegra2 failures?

Tom
Simon Glass - Feb. 24, 2012, 5:57 p.m.
On Thu, Feb 23, 2012 at 1:25 PM, Tom Warren <twarren.nvidia@gmail.com> wrote:
> On Fri, Feb 17, 2012 at 9:01 AM, Tom Warren <twarren.nvidia@gmail.com> wrote:
>> The 4.2.2 gcc in the ELDK42 release doesn't like the direct SP
>> load using a constant in tegra2_start. Change it to use a load
>> thru another reg using mov sp, %0 : : "r"(CONST).
>>
>> Tested on my Seaboard T20-A03, U-Boot loads and runs OK. Also
>> compiled all tegra2 builds with both gcc 4.2.2 and 4.4.1 OK.
>>
>> Signed-off-by: Tom Warren <twarren@nvidia.com>

Tested on gcc 4.6, sorry I don't have the compiler that fails.

Acked-by: Simon Glass <sjg@chromium.org>
Tested-by: Simon Glass <sjg@chromium.org>

>> ---
>> V2: Change 4 mov/orr's to single mov sp as per Marek Vasut
>>
>>  arch/arm/cpu/armv7/tegra2/ap20.c |   10 +++++-----
>>  1 files changed, 5 insertions(+), 5 deletions(-)
>>
>> diff --git a/arch/arm/cpu/armv7/tegra2/ap20.c b/arch/arm/cpu/armv7/tegra2/ap20.c
>> index 4c44bb3..f0450f4 100644
>> --- a/arch/arm/cpu/armv7/tegra2/ap20.c
>> +++ b/arch/arm/cpu/armv7/tegra2/ap20.c
>> @@ -298,11 +298,11 @@ void tegra2_start(void)
>>                writel(0xC0, &pmt->pmt_cfg_ctl);
>>
>>                /*
>> -               * If we are ARM7 - give it a different stack. We are about to
>> -               * start up the A9 which will want to use this one.
>> -               */
>> -               asm volatile("ldr       sp, =%c0\n"
>> -                       : : "i"(AVP_EARLY_BOOT_STACK_LIMIT));
>> +                * If we are ARM7 - give it a different stack. We are about to
>> +                * start up the A9 which will want to use this one.
>> +                */
>> +               asm volatile("mov       sp, %0\n"
>> +                       : : "r"(AVP_EARLY_BOOT_STACK_LIMIT));
>>
>>                start_cpu((u32)_start);
>>                halt_avp();
>> --
>> 1.7.7.1
>>
>
> Ack? Nak? Can I get some love here?
>
> Albert/Wolfgang - did anyone try this to ensure that it really fixes
> the 6 Tegra2 failures?
>
> Tom
> _______________________________________________
> U-Boot mailing list
> U-Boot@lists.denx.de
> http://lists.denx.de/mailman/listinfo/u-boot

Patch

diff --git a/arch/arm/cpu/armv7/tegra2/ap20.c b/arch/arm/cpu/armv7/tegra2/ap20.c
index 4c44bb3..f0450f4 100644
--- a/arch/arm/cpu/armv7/tegra2/ap20.c
+++ b/arch/arm/cpu/armv7/tegra2/ap20.c
@@ -298,11 +298,11 @@  void tegra2_start(void)
 		writel(0xC0, &pmt->pmt_cfg_ctl);
 
 		/*
-		* If we are ARM7 - give it a different stack. We are about to
-		* start up the A9 which will want to use this one.
-		*/
-		asm volatile("ldr	sp, =%c0\n"
-			: : "i"(AVP_EARLY_BOOT_STACK_LIMIT));
+		 * If we are ARM7 - give it a different stack. We are about to
+		 * start up the A9 which will want to use this one.
+		 */
+		asm volatile("mov	sp, %0\n"
+			: : "r"(AVP_EARLY_BOOT_STACK_LIMIT));
 
 		start_cpu((u32)_start);
 		halt_avp();