Skip to contentSkip to conversation

powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

PATCH New Timur Tabi 5 participants 7 messages Mar 17, 2012 – Jun 15, 2012

lore
Share & tools
lore

Timur Tabi

author
#0 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

The malloc buffer is not large enough to hold a flash sector (0x20000 bytes) in addition to whatever else it normally holds, so double its size. This fixes a failure trying to save the environment:

=> save
Saving Environment to Flash...
Unable to save the rest of sector (122880)
. done
Protected 1 sectors

This problem probably surfaced from some other change that significantly increased the normal memory usage, thereby not leaving enough room for the saveenv command.

Signed-off-by Timur Tabi <[email protected]>
---
 include/configs/MPC832XEMDS.h |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
include/configs/MPC832XEMDS.h +1 -1
index 4ed5a97..6f8622c 100644
@@ -181,7 +181,7 @@
 
 /* CONFIG_SYS_MONITOR_LEN must be a multiple of CONFIG_ENV_SECT_SIZE */
 #define CONFIG_SYS_MONITOR_LEN	(384 * 1024)	/* Reserve 384 kB for Mon */
-#define CONFIG_SYS_MALLOC_LEN	(128 * 1024)	/* Reserved for malloc */
+#define CONFIG_SYS_MALLOC_LEN	(256 * 1024)	/* Reserved for malloc */
 
 /*
  * Initial RAM Base Address Setup
Signature · 1 line
1.7.4.4
#1 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

Dear Timur Tabi,

Quoted 3 lines · In message <[email protected]> you wrote:

The malloc buffer is not large enough to hold a flash sector (0x20000 bytes) in addition to whatever else it normally holds, so double its size. This fixes a failure trying to save the environment:

Doubling it is kind of aggressive strategy. You know exactly how much free room needs to be guaranteed, so why don't you auto-adjust the size?

include/configs/MPC832XEMDS.h line 184

Quoted 2 lines
include/configs/MPC832XEMDS.h +1 -1
-#define CONFIG_SYS_MALLOC_LEN	(128 * 1024)	/* Reserved for malloc */
+#define CONFIG_SYS_MALLOC_LEN	(256 * 1024)	/* Reserved for malloc */

How about:

#define CONFIG_SYS_MALLOC_LEN (128 * 1024 + CONFIG_ENV_SECT_SIZE)

?

Yes, your board config file is kind of broken and will need more cleanup to facilitate this, but that should be done anyway.

Why exactly don't you support NOR flash for RAM-booting configurations?

Best regards,

Wolfgang Denk

Signature · 7 lines
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 at denx.de
The evolution of the human race will not be accomplished in  the  ten
thousand  years  of  tame  animals,  but in the million years of wild
animals, because man is and will always be a wild animal.
                                              - Charles Galton Darwin
#2 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

Quoted 10 lines · Wolfgang Denk

Doubling it is kind of aggressive strategy. ?You know exactly how much free room needs to be guaranteed, so why don't you auto-adjust the size?

Review

Quoted 2 lines
(quoted diff) +1 -1
-#define CONFIG_SYS_MALLOC_LEN ? ? ? ?(128 * 1024) ? ?/* Reserved for malloc */
+#define CONFIG_SYS_MALLOC_LEN ? ? ? ?(256 * 1024) ? ?/* Reserved for malloc */

How about:

#define CONFIG_SYS_MALLOC_LEN ? (128 * 1024 + CONFIG_ENV_SECT_SIZE)

That's the same thing. CONFIG_ENV_SECT_SIZE is 128KB. Plus, as you realized, this will not work if CONFIG_SYS_RAMBOOT is defined.

Quoted 2 lines

Yes, your board config file is kind of broken and will need more cleanup to facilitate this, but that should be done anyway.

Well, I have no desire to do that. I'm just fixing one bug. A major cleanup of the board file is not on my to-do list.

Quoted 1 line

Why exactly don't you support NOR flash for RAM-booting configurations?

I have no idea. I just noticed this one bug. I don't support the board.

Signature · 2 lines
Timur Tabi
Linux kernel developer at Freescale
#3 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

Dear Tabi Timur-B04825,

Quoted 7 lines · In message <CAOZdJXUem_-mgDSqF+p4R0npZQ6qu14G1yexQCBUe2G=nr2-1Q@mail.gmail.com>…
Quoted 6 lines

Doubling it is kind of aggressive strategy. You know exactly how much free room needs to be guaranteed, so why don't you auto-adjust the size?

Quoted 1 line

-#define CONFIG_SYS_MALLOC_LEN (128 * 1024) /* Res=

erved for malloc */

Quoted 8 lines
Quoted 1 line
Quoted 1 line

+#define CONFIG_SYS_MALLOC_LEN (256 * 1024) /* Res=

erved for malloc */

Quoted 4 lines

How about:

#define CONFIG_SYS_MALLOC_LEN (128 * 1024 + CONFIG_ENV_SECT_SIZE)

That's the same thing. CONFIG_ENV_SECT_SIZE is 128KB. Plus, as you

No, it's not the same thing. Your code is static an either wasted moemory on systems with smaller sector size, or needs to be readjusted on systems with bigger sectors.

Don;t just think about your current situation, but also what happens when you (or somebody else) copies that code for other systems.

Make your code more robust.

Quoted 1 line

realized, this will not work if CONFIG_SYS_RAMBOOT is defined.

I consider this a bug that needs fixing anyway.

Quoted 5 lines
Quoted 2 lines

Yes, your board config file is kind of broken and will need more cleanup to facilitate this, but that should be done anyway.

Well, I have no desire to do that. I'm just fixing one bug. A major cleanup of the board file is not on my to-do list.

Well, fixing this bug has uncovered another shortcoming or bug (depending on point of view) of the code, so we should fix that one, too.

Best regards,

Wolfgang Denk

Signature · 5 lines
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 at denx.de
The goal of science is to build better mousetraps. The goal of nature
is to build better mice.
#4 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

Quoted 13 lines · Wolfgang Denk

Dear Tabi Timur-B04825,

Quoted 10 lines · In message <CAOZdJXUem_-mgDSqF+p4R0npZQ6qu14G1yexQCBUe2G=nr2-1Q@mail.gmail.com>…
Quoted 6 lines

Doubling it is kind of aggressive strategy. You know exactly how much free room needs to be guaranteed, so why don't you auto-adjust the size?

Quoted 1 line

-#define CONFIG_SYS_MALLOC_LEN (128 * 1024) /* Res=

erved for malloc */

Quoted 1 line
Quoted 1 line

+#define CONFIG_SYS_MALLOC_LEN (256 * 1024) /* Res=

erved for malloc */

Quoted 1 line
Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

How about:

#define CONFIG_SYS_MALLOC_LEN (128 * 1024 + CONFIG_ENV_SECT_SIZE)

That's the same thing. CONFIG_ENV_SECT_SIZE is 128KB. Plus, as you

No, it's not the same thing.

You really want CONFIG_SYS_MALLOC_LEN to be a list of everything that could be mallocked -- or worse, a list of just the things that people have noticed breaking in the past (even though other things could later be depending on that increase in size)? You want to answer questions about why NAND breaks on a board with a smaller NOR sector size? :-P

Quoted 1 line

Your code is static an either wasted moemory on systems with smaller sector size,

Come on, you're worried about a 128K increase in memory reserved for the use of the bootloader? It's not "wasted". The only thing that area of memory could otherwise be used for is loading images and such, and it is very unlikely that this small change is going to matter there.

Quoted 6 lines

or needs to be readjusted on systems with bigger sectors.

Don;t just think about your current situation, but also what happens when you (or somebody else) copies that code for other systems.

Make your code more robust.

The robust thing to do would be to not be stingy with the malloc size, and change all boards to have at least 1 MiB for malloc (except ones with very small amounts of RAM). Maybe have a debug command to print max memory usage since boot, to see if a board is getting anywhere near the limit.

-Scott

#5 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

Dear Scott Wood,

Quoted 8 lines · In message <[email protected]> you wrote:
Quoted 2 lines

Make your code more robust.

The robust thing to do would be to not be stingy with the malloc size, and change all boards to have at least 1 MiB for malloc (except ones with very small amounts of RAM). Maybe have a debug command to print max memory usage since boot, to see if a board is getting anywhere near the limit.

Good idea. Patches are welcome, as usual.

I'm not sure it the malloc code keeps any such statics - if yes, and this can be done without any significant amount of code, this could even be added to the default bdinfo output.

Best regards,

Wolfgang Denk

Signature · 5 lines
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 at denx.de
Reader, suppose you were an idiot. And suppose you were a  member  of
Congress. But I repeat myself.                           - Mark Twain
Applied
#6 lore raw

PATCH 1/1 powerpc/83xx: increment malloc heap size for the MPC832x MDS boards

In reply to Timur Tabi
Quoted 10 lines · On Sat, 17 Mar 2012 17:44:00 -0500 Timur Tabi <[email protected]> wrote:

The malloc buffer is not large enough to hold a flash sector (0x20000 bytes) in addition to whatever else it normally holds, so double its size. This fixes a failure trying to save the environment:

=> save
Saving Environment to Flash...
Unable to save the rest of sector (122880)
. done
Protected 1 sectors

Quoted 6 lines

This problem probably surfaced from some other change that significantly increased the normal memory usage, thereby not leaving enough room for the saveenv command.

Signed-off-by Timur Tabi <[email protected]>

---

applied, Thanks.

Kim