From: Marek Vasut <marex@denx.de>
To: u-boot@lists.denx.de
Subject: [U-Boot] [PATCH v3 2/2] arm: add initial support for Amlogic Meson and ODROID-C2
Date: Wed, 13 Apr 2016 00:26:43 +0200 [thread overview]
Message-ID: <570D7623.8020603@denx.de> (raw)
In-Reply-To: <20160412214945.GA3006@gmail.com>
On 04/12/2016 11:50 PM, Beniamino Galvani wrote:
> On Mon, Apr 11, 2016 at 11:08:02PM +0200, Marek Vasut wrote:
>>>>> + val = fdt_getprop(gd->fdt_blob, offset, "reg", &len);
>>>>> + if (len < sizeof(*val) * 4)
>>>>> + return -EINVAL;
>>>>> +
>>>>> + /* Don't use fdt64_t to avoid unaligned access */
>>>>
>>>> This looks iffy, can you elaborate on this issue ?
>>>
>>> I was getting a "Synchronous Abort handler, esr 0x96000021" which
>>> seemed to indicate a alignment fault, but thinking again about it I'm
>>> not sure anymore of the real cause. fdt64_t and fdt64_to_cpu() don't
>>> work here, I will try to investigate better why. Suggestions are
>>> welcome :)
>>
>> Toolchain issues ? Stack alignment issue ?
>
> So, after some investigation, the reason is that the code runs when
> caches are still disabled and thus all the memory is treated as
> Device-nGnRnE, requiring aligned accesses.
You mean 8-byte aligned accesses, correct ?
> The return value of
> fdt_getprop() is guaranteed to be aligned to a 4 byte boundary (but
> not 8)
The return value of fdt_getprop() is a pointer, thus 8byte long on
aarch64 and thus aligned to 8 bytes on the stack unless there is
some real problem.
> and therefore a 32-bit type must be used to avoid alignment
> faults. Probably the comment should be updated to explain this better.
Take a look at what uniphier does : arch/arm/mach-uniphier/dram_init.c
Does that approach with fdt64_t work for you?
> Beniamino
>
Code in question:
+int dram_init(void)
+{
+ const fdt32_t *val;
+ int offset;
+ int len;
+
+ offset = fdt_path_offset(gd->fdt_blob, "/memory");
+ if (offset < 0)
+ return -EINVAL;
+
+ val = fdt_getprop(gd->fdt_blob, offset, "reg", &len);
+ if (len < sizeof(*val) * 4)
+ return -EINVAL;
+
+ /* Don't use fdt64_t to avoid unaligned access */
+ gd->ram_size = (uint64_t)fdt32_to_cpu(val[2]) << 32;
+ gd->ram_size |= fdt32_to_cpu(val[3]);
+
+ return 0;
+}
--
Best regards,
Marek Vasut
next prev parent reply other threads:[~2016-04-12 22:26 UTC|newest]
Thread overview: 19+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-04-10 15:30 [U-Boot] [PATCH v3 0/2] Amlogic Meson GXBaby and ODROID-C2 support Beniamino Galvani
2016-04-10 15:30 ` [U-Boot] [PATCH v3 1/2] net: designware: fix descriptor layout and warnings on 64-bit archs Beniamino Galvani
2016-04-10 23:59 ` Marek Vasut
2016-04-11 20:46 ` Beniamino Galvani
2016-04-10 15:30 ` [U-Boot] [PATCH v3 2/2] arm: add initial support for Amlogic Meson and ODROID-C2 Beniamino Galvani
2016-04-11 0:08 ` Marek Vasut
2016-04-11 21:02 ` Beniamino Galvani
2016-04-11 21:08 ` Marek Vasut
2016-04-12 21:50 ` Beniamino Galvani
2016-04-12 22:26 ` Marek Vasut [this message]
2016-04-13 11:22 ` Beniamino Galvani
2016-04-13 11:52 ` Marek Vasut
2016-04-13 21:38 ` Alexander Graf
2016-04-13 22:34 ` Tom Rini
2016-04-13 22:42 ` Alexander Graf
2016-04-13 22:51 ` Tom Rini
2016-04-13 22:53 ` Alexander Graf
2016-04-13 23:12 ` Marek Vasut
2016-04-14 7:08 ` Alexander Graf
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=570D7623.8020603@denx.de \
--to=marex@denx.de \
--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