All of lore.kernel.org
 help / color / mirror / Atom feed
From: Albert ARIBAUD <albert.u.boot@aribaud.net>
To: u-boot@lists.denx.de
Subject: [U-Boot] [PATCH v2] config.mk: use memoization in cc-option macro to speed up compilation
Date: Sat, 05 Nov 2011 16:02:45 +0100	[thread overview]
Message-ID: <4EB55015.9080903@aribaud.net> (raw)
In-Reply-To: <4EB53D79.7090207@googlemail.com>

Hi Daniel,

Le 05/11/2011 14:43, Daniel Schwierzeck a ?crit :
> Hi Albert,
>
> On 05.11.2011 10:16, Albert ARIBAUD wrote:
>> Hi all,
>>
>> Le 04/11/2011 18:56, Wolfgang Denk a ?crit :
>>> Dear Daniel Schwierzeck,
>>>
>>> In
>>> message<CACUy__XY1873+dT-s=-GyPeBq5hnYsmYwWt-gxEXiwY97EcGdg@mail.gmail.com>
>>>
>>> you wrote:
>>>>
>>>> Should we change it? is the semantic still the same?
>>>
>>> I'm not sure. At first reading it doesn't look really the same to me.
>>
>> They are not, at least for ELDK4.2.
>>
>> The only difference is in -mabi options, where the change would reduce
>> "-mabi=apcs-gnu -mabi=aapcs-linux" to "-mabi=aapcs-linux".
>>
>> apcs-gnu, IIUC, is 'old ABI', while 'aapcs-linux' is 'new ABI', aka
>> eabi. Most of the toolchains I see are eabi (ELDK and CS notably). There
>> may be 'old ABI' toolchains out there, but I don't think they are old
>> ABI either.
>>
>> Anyway, I've just tried ./MAKEALL edminiv2 with ELD42 and a couple of CS
>> toolchains, and nowhere in the log does -mabi=apcs-gnu show up -- the
>> gcc invocations only have -mabi=aapcs-linux.
>
> Looks like I read it wrong. So you always want "-mabi=apcs-gnu
> -mabi=aapcs-linux -mno-thumb-interwork" in $(PF_CPPFLAGS_ABI) with EABI?
> Sorry but I am not an ARM expert ;)

No, I don't want that. :)

The problem I see is having two conflicting -mabi options, 
-mabi=apcs-gnu and -mabi=aapcs-linux, in the same command line. There 
should be only one -- and it should be the same across the whole U-Boot 
building process.

The duplicate -mno-thumb-interwork does not worry me fronm a functional 
standpoint; it's just a waste of space, that's all.

>> I've also tested making ED Mini V2 with and without the patch but
>> without Daniel's proposed change to arch/arm/config.mk, and there is no
>> difference in build commands (except that for some reason the patch
>> inserts multiple spaces between some gcc invocation options.
>
> I guess this comes from making the macros more readable. Maybe I can
> optimize this.
>
>>
>> Daniel, what do you mean with "does not work correctly"?
>
> that the generated cache file looks not right
>
> CC_OPTIONS += -mabi=apcs-gnu
> CC_OPTIONS_NOP += -mabi=apcs-gnu
> CC_OPTIONS += -mno-thumb-interwork
> CC_OPTIONS += -mabi=aapcs-linux -mno-thumb-interwork

I'm not a makefile expert, so "looks not right" is a bit meaningless to 
me. If "not right" means "there are several conflicting -mabi options 
and there are repeated -mno-thumb-interwork options" then I agree with 
your "does not look right" statement.

> But if you want "-mabi=apcs-gnu -mabi=aapcs-linux -mno-thumb-interwork"
> then it should already work correctly without my change in
> arch/arm/config.mk.

As I said, no, I don't want that. I want the right -mabi option only, 
and I'd like a single -mno-thumb-interwork option.

But what I don't understand is the discrepancy which I see between the 
CC_OPTIONS resulting from include/generated/cc_options.mk and the actual 
command line options used to generate e.g. edminiv2.

> Best regards,
> Daniel

Amicalement,
-- 
Albert.

  reply	other threads:[~2011-11-05 15:02 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-11-02 18:46 [U-Boot] [PATCH] config.mk: use memoization in cc-option macro to speed up compilation Daniel Schwierzeck
2011-11-02 20:39 ` Simon Glass
2011-11-02 21:36   ` Daniel Schwierzeck
2011-11-03 19:47     ` Wolfgang Denk
2011-11-04 12:53 ` [U-Boot] [PATCH v2] " Daniel Schwierzeck
2011-11-04 16:32   ` Simon Glass
2011-11-04 17:15     ` Daniel Schwierzeck
2011-11-04 17:31   ` Daniel Schwierzeck
2011-11-04 17:56     ` Wolfgang Denk
2011-11-05  9:16       ` Albert ARIBAUD
2011-11-05 13:30         ` Wolfgang Denk
2011-11-05 13:43         ` Daniel Schwierzeck
2011-11-05 15:02           ` Albert ARIBAUD [this message]
2011-11-07 15:26   ` [U-Boot] [PATCH v3] " Daniel Schwierzeck
2011-11-07 21:07     ` Wolfgang Denk
     [not found] ` <1320410756-30391-1-git-send-email-daniel.schwierzeck@googlemail.com>
2011-11-05  7:17   ` [U-Boot] [PATCH v2] " Aneesh V

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=4EB55015.9080903@aribaud.net \
    --to=albert.u.boot@aribaud.net \
    --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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.