[U-Boot] [PATCH] mtest: Disable dcache during test

[U-Boot] [PATCH] mtest: Disable dcache during test

[Benoît Thébaudeau]

On 08/11/2012 21:47, Andrew Dyer wrote:
> I agree with Mike, use the current dcache settings. U-boot has always
> assumed the user knew what they were doing.
> If you want to print a small message with the dcache setting that
> makes sense to me, but no big warning.

Then, something like the following at runtime in the first lines printed my
mtest?

dcache state: on

Best regards,
Beno?t

[U-Boot] [PATCH] mtest: Disable dcache during test

[Mike Frysinger]

On Saturday 11 August 2012 16:05:36 Beno?t Th?baudeau wrote:
> On 08/11/2012 21:47, Andrew Dyer wrote:
> > I agree with Mike, use the current dcache settings. U-boot has always
> > assumed the user knew what they were doing.
> > If you want to print a small message with the dcache setting that
> > makes sense to me, but no big warning.
> 
> Then, something like the following at runtime in the first lines printed my
> mtest?
> 
> dcache state: on

i'd be fine with that.  something like below (if you want to test & post,
that'd be good).
-mike

diff --git a/common/cmd_mem.c b/common/cmd_mem.c
index 18f0a3f..5628f6a 100644
--- a/common/cmd_mem.c
+++ b/common/cmd_mem.c
@@ -651,8 +651,10 @@ int do_mem_mtest (cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	else
 		iteration_limit = 0;
 
+	printf("Testing %08x ... %08x (dcache: %s):\n", (uint)start, (uint)end,
+		dcache_status() ? "on" : "off");
+
 #if defined(CONFIG_SYS_ALT_MEMTEST)
-	printf ("Testing %08x ... %08x:\n", (uint)start, (uint)end);
 	debug("%s:%d: start 0x%p end 0x%p\n",
 		__FUNCTION__, __LINE__, start, end);
 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20120811/5c64e75b/attachment.pgp>

[U-Boot] [PATCH] mtest: Disable dcache during test

[Benoît Thébaudeau]

On Saturday 11 August 2012 22:15:16 Mike Frysinger wrote:
> On Saturday 11 August 2012 16:05:36 Beno?t Th?baudeau wrote:
> > On 08/11/2012 21:47, Andrew Dyer wrote:
> > > I agree with Mike, use the current dcache settings. U-boot has
> > > always
> > > assumed the user knew what they were doing.
> > > If you want to print a small message with the dcache setting that
> > > makes sense to me, but no big warning.
> > 
> > Then, something like the following at runtime in the first lines
> > printed my
> > mtest?
> > 
> > dcache state: on
> 
> i'd be fine with that.  something like below (if you want to test &
> post,
> that'd be good).
> -mike
> 
> diff --git a/common/cmd_mem.c b/common/cmd_mem.c
> index 18f0a3f..5628f6a 100644
> --- a/common/cmd_mem.c
> +++ b/common/cmd_mem.c
> @@ -651,8 +651,10 @@ int do_mem_mtest (cmd_tbl_t *cmdtp, int flag,
> int argc, char * const argv[])
>  	else
>  		iteration_limit = 0;
>  
> +	printf("Testing %08x ... %08x (dcache: %s):\n", (uint)start,
> (uint)end,
> +		dcache_status() ? "on" : "off");
> +
>  #if defined(CONFIG_SYS_ALT_MEMTEST)
> -	printf ("Testing %08x ... %08x:\n", (uint)start, (uint)end);
>  	debug("%s:%d: start 0x%p end 0x%p\n",
>  		__FUNCTION__, __LINE__, start, end);

OK, I'll do that next week.

Best regards,
Beno?t

[U-Boot] [PATCH] mtest: Print dcache state

[Benoît Thébaudeau]

mtest tests many types of memory accesses in many different conditions. If
dcache is enabled, memory accesses are likely bursts, and some memory accesses
are simply skipped. Hence, mtest results may change depending on dcache state.

This patch prints the dcache state at the beginning of mtest so as to keep the
user informed of the test conditions.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Tested-by: Beno?t Th?baudeau <benoit.thebaudeau@advansee.com>
Cc: Wolfgang Denk <wd@denx.de>
Cc: Mike Frysinger <vapier@gentoo.org>
---
 .../common/cmd_mem.c                               |    4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git u-boot-4d3c95f.orig/common/cmd_mem.c u-boot-4d3c95f/common/cmd_mem.c
index 18f0a3f..5628f6a 100644
--- u-boot-4d3c95f.orig/common/cmd_mem.c
+++ u-boot-4d3c95f/common/cmd_mem.c
@@ -651,8 +651,10 @@ int do_mem_mtest (cmd_tbl_t *cmdtp, int flag, int argc, char * const argv[])
 	else
 		iteration_limit = 0;
 
+	printf("Testing %08x ... %08x (dcache: %s):\n", (uint)start, (uint)end,
+		dcache_status() ? "on" : "off");
+
 #if defined(CONFIG_SYS_ALT_MEMTEST)
-	printf ("Testing %08x ... %08x:\n", (uint)start, (uint)end);
 	debug("%s:%d: start 0x%p end 0x%p\n",
 		__FUNCTION__, __LINE__, start, end);
 

[U-Boot] mtest: Print dcache state

[Tom Rini]

On Mon, Aug 13, 2012 at 12:59:12AM -0000, Beno??t Th??baudeau wrote:

> mtest tests many types of memory accesses in many different conditions. If
> dcache is enabled, memory accesses are likely bursts, and some memory accesses
> are simply skipped. Hence, mtest results may change depending on dcache state.
> 
> This patch prints the dcache state at the beginning of mtest so as to keep the
> user informed of the test conditions.
> 
> Signed-off-by: Mike Frysinger <vapier@gentoo.org>
> Tested-by: Beno??t Th??baudeau <benoit.thebaudeau@advansee.com>
> Cc: Wolfgang Denk <wd@denx.de>
> Cc: Mike Frysinger <vapier@gentoo.org>

NAK, this breaks x86 which has no dcache_status.

-- 
Tom
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 836 bytes
Desc: Digital signature
URL: <http://lists.denx.de/pipermail/u-boot/attachments/20121003/0ef4b6fb/attachment.pgp>

[U-Boot] mtest: Print dcache state

[Albert ARIBAUD]

Hi Tom,

On Wed, 3 Oct 2012 15:05:21 -0700, Tom Rini <trini@ti.com> wrote:

> On Mon, Aug 13, 2012 at 12:59:12AM -0000, Beno??t Th??baudeau wrote:
> 
> > mtest tests many types of memory accesses in many different conditions. If
> > dcache is enabled, memory accesses are likely bursts, and some memory accesses
> > are simply skipped. Hence, mtest results may change depending on dcache state.
> > 
> > This patch prints the dcache state at the beginning of mtest so as to keep the
> > user informed of the test conditions.
> > 
> > Signed-off-by: Mike Frysinger <vapier@gentoo.org>
> > Tested-by: Beno??t Th??baudeau <benoit.thebaudeau@advansee.com>
> > Cc: Wolfgang Denk <wd@denx.de>
> > Cc: Mike Frysinger <vapier@gentoo.org>
> 
> NAK, this breaks x86 which has no dcache_status.
> 

IIRC the general idea of printing the cache status in mtest had already
been NAKed (though not formally) by Wolfgang.

Amicalement,
-- 
Albert.

[U-Boot] mtest: Print dcache state

[Benoît Thébaudeau]

Hi Albert, Tom,

On Thursday, October 4, 2012 1:03:48 PM, Albert ARIBAUD wrote:
> Hi Tom,
> 
> On Wed, 3 Oct 2012 15:05:21 -0700, Tom Rini <trini@ti.com> wrote:
> 
> > On Mon, Aug 13, 2012 at 12:59:12AM -0000, Beno??t Th??baudeau
> > wrote:
> > 
> > > mtest tests many types of memory accesses in many different
> > > conditions. If
> > > dcache is enabled, memory accesses are likely bursts, and some
> > > memory accesses
> > > are simply skipped. Hence, mtest results may change depending on
> > > dcache state.
> > > 
> > > This patch prints the dcache state at the beginning of mtest so
> > > as to keep the
> > > user informed of the test conditions.
> > > 
> > > Signed-off-by: Mike Frysinger <vapier@gentoo.org>
> > > Tested-by: Beno??t Th??baudeau <benoit.thebaudeau@advansee.com>
> > > Cc: Wolfgang Denk <wd@denx.de>
> > > Cc: Mike Frysinger <vapier@gentoo.org>
> > 
> > NAK, this breaks x86 which has no dcache_status.
> > 
> 
> IIRC the general idea of printing the cache status in mtest had
> already
> been NAKed (though not formally) by Wolfgang.

Indeed. This subject has already been closed. Perhaps the patch status had just
not been updated in Patchwork.

Best regards,
Beno?t

[U-Boot] mtest: Print dcache state

[Tom Rini]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 10/04/12 04:13, Beno?t Th?baudeau wrote:
> Hi Albert, Tom,
> 
> On Thursday, October 4, 2012 1:03:48 PM, Albert ARIBAUD wrote:
>> Hi Tom,
>> 
>> On Wed, 3 Oct 2012 15:05:21 -0700, Tom Rini <trini@ti.com>
>> wrote:
>> 
>>> On Mon, Aug 13, 2012 at 12:59:12AM -0000, Beno??t Th??baudeau 
>>> wrote:
>>> 
>>>> mtest tests many types of memory accesses in many different 
>>>> conditions. If dcache is enabled, memory accesses are likely
>>>> bursts, and some memory accesses are simply skipped. Hence,
>>>> mtest results may change depending on dcache state.
>>>> 
>>>> This patch prints the dcache state at the beginning of mtest
>>>> so as to keep the user informed of the test conditions.
>>>> 
>>>> Signed-off-by: Mike Frysinger <vapier@gentoo.org> Tested-by:
>>>> Beno??t Th??baudeau <benoit.thebaudeau@advansee.com> Cc:
>>>> Wolfgang Denk <wd@denx.de> Cc: Mike Frysinger
>>>> <vapier@gentoo.org>
>>> 
>>> NAK, this breaks x86 which has no dcache_status.
>>> 
>> 
>> IIRC the general idea of printing the cache status in mtest had 
>> already been NAKed (though not formally) by Wolfgang.
> 
> Indeed. This subject has already been closed. Perhaps the patch
> status had just not been updated in Patchwork.

OK, thanks guys.

- -- 
Tom
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://www.enigmail.net/
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=OpC9
-----END PGP SIGNATURE-----

[U-Boot] [PATCH] mtest: Disable dcache during test

[Albert ARIBAUD]

Hi Mike,

On Sat, 11 Aug 2012 16:15:16 -0400, Mike Frysinger <vapier@gentoo.org>
wrote:
> On Saturday 11 August 2012 16:05:36 Beno?t Th?baudeau wrote:
> > On 08/11/2012 21:47, Andrew Dyer wrote:
> > > I agree with Mike, use the current dcache settings. U-boot has
> > > always assumed the user knew what they were doing.
> > > If you want to print a small message with the dcache setting that
> > > makes sense to me, but no big warning.
> > 
> > Then, something like the following at runtime in the first lines
> > printed my mtest?
> > 
> > dcache state: on
> 
> i'd be fine with that.  something like below (if you want to test &
> post, that'd be good).
> -mike

I'd disagree with this solution; one can easily overlook casual test
messages like these. To me it is better that the person doing mtest not
trust the cache state at all and check / set it.

Amicalement,
-- 
Albert.