From: "André Przywara" <andre.przywara@arm.com>
To: u-boot@lists.denx.de
Subject: [U-Boot] [PATCH v2] SPL: tiny-printf: avoid any BSS usage
Date: Sat, 9 Jul 2016 18:27:41 +0100 [thread overview]
Message-ID: <07b8317d-ddb2-c914-f85c-9b56e2a7e01d@arm.com> (raw)
In-Reply-To: <CAPnjgZ36VFePkp-ZVALBqWZQVs50=dYTiKSe-Xj1KsaRK78iSw@mail.gmail.com>
On 09/07/16 15:38, Simon Glass wrote:
> On 8 July 2016 at 08:18, Andre Przywara <andre.przywara@arm.com> wrote:
>> As printf calls may be executed quite early, we should avoid using any
>> BSS stored variables, since some boards put BSS in DRAM, which may not
>> have been initialised yet.
>> Explicitly mark those "static global" variables as belonging to the
>> .data section, to keep tiny-printf clear of any BSS usage.
>>
>> Signed-off-by: Andre Przywara <andre.przywara@arm.com>
>
> Reviewed-by: Simon Glass <sjg@chromium.org>
>
> Another approach would be to put these vars in a struct, declare it as
> a local variable and pass it around. But this works OK too.
Yeah, I was thinking about this too. Actually fixing the issue that we
need global variables in the first place.
But a quick look revealed that this is not trivial, so I reverted to
this simpler approach for the quick fix. Also I am not sure this will
eventually work against this "tiny" idea, we will see.
To be honest I think the real proper fix(TM) would be to provide a
separate BSS section for any code that runs _before_ DRAM init. This
should be a few bytes only and could easily live in SRAM.
If the loading part of the SPL requires a bigger BSS, that's fine as it
could still live in DRAM.
I might bake a patch when I get bored ...
Cheers,
Andre.
>
>> ---
>> Changes v1 .. v2:
>> - remove NO_BSS macro, use __attribute__ ... directly instead
>>
>> lib/tiny-printf.c | 11 ++++++++---
>> 1 file changed, 8 insertions(+), 3 deletions(-)
>>
>> diff --git a/lib/tiny-printf.c b/lib/tiny-printf.c
>> index 451f4f7..b334f05 100644
>> --- a/lib/tiny-printf.c
>> +++ b/lib/tiny-printf.c
>> @@ -13,11 +13,16 @@
>> #include <stdarg.h>
>> #include <serial.h>
>>
>> -static char *bf;
>> -static char zs;
>> +/*
>> + * This code in here may execute before the DRAM is initialised, so
>> + * we should make sure that it doesn't touch BSS, which some boards
>> + * put in DRAM.
>> + */
>> +static char *bf __attribute__ ((section(".data")));
>> +static char zs __attribute__ ((section(".data")));
>>
>> /* Current position in sprintf() output string */
>> -static char *outstr;
>> +static char *outstr __attribute__ ((section(".data")));
>>
>> static void out(char c)
>> {
>> --
>> 2.8.2
>>
>
next prev parent reply other threads:[~2016-07-09 17:27 UTC|newest]
Thread overview: 10+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-07-08 14:18 [U-Boot] [PATCH v2] SPL: tiny-printf: avoid any BSS usage Andre Przywara
2016-07-08 16:50 ` [U-Boot] [U-Boot,v2] " Tom Rini
2016-07-08 21:48 ` Marek Vasut
2016-07-09 0:28 ` Tom Rini
2016-07-09 8:22 ` Marek Vasut
2016-07-09 14:50 ` Tom Rini
2016-07-09 14:38 ` [U-Boot] [PATCH v2] " Simon Glass
2016-07-09 15:16 ` Marek Vasut
2016-07-09 17:27 ` André Przywara [this message]
2016-08-06 0:11 ` Simon Glass
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=07b8317d-ddb2-c914-f85c-9b56e2a7e01d@arm.com \
--to=andre.przywara@arm.com \
--cc=u-boot@lists.denx.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox