linux-arm-kernel.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: u.kleine-koenig@pengutronix.de (Uwe Kleine-König)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH] clk: don't use __initconst for non-const arrays
Date: Fri, 12 Sep 2014 09:59:52 +0200	[thread overview]
Message-ID: <20140912075952.GR3755@pengutronix.de> (raw)
In-Reply-To: <CAKv+Gu9LubyPHoYJsRfRoqgONAApb3nEoXAxUa9g6tecP=McOw@mail.gmail.com>

Hello Ard,

On Fri, Sep 12, 2014 at 09:42:29AM +0200, Ard Biesheuvel wrote:
> On 12 September 2014 00:04, Uwe Kleine-K?nig
> <u.kleine-koenig@pengutronix.de> wrote:
> > Hello,
> >
> > On Thu, Sep 11, 2014 at 11:04:31PM +0200, Uwe Kleine-K?nig wrote:
> >>  /* Mux parent lists. */
> >> -static const char *fin_pll_p[] __initconst = {
> >> +static const char *fin_pll_p[] __initdata = {
> >>       "xxti",
> >>       "xusbxti"
> >>  };
> > As discussed with Tomasz on irc: The sad thing here is that for this
> > array only 8 bytes are freed when .init.rodata is thrown away (that is
> > two pointers). The actual data---5 + 8 bytes + maybe aligning---isn't
> > freed though.
> >
> > We wondered if there is a nice and easy way to throw away the
> > characters, too.
> >
> > The only way I currently see is:
> >
> >         const char xxti[] __initconst = "xxti";
> >         ...
> >
> >         static const char *fin_pll_p[] __initdata = {
> >                 xxti,
> >                 ...
> >         };
> >
> > but this definitively doesn't qualify as "nice and easy". Is there an
> > alternative?
> >
> 
> What about doing
> 
> static const char fin_pll_p[][8] __initconst = {
>      "xxti",
>      "xusbxti"
> };
This results in the strings being moved to .init.rodata and so they are
discarded. But it also results in:

	drivers/clk/samsung/clk-s5pv210.c:412:2: warning: initialization from incompatible pointer type [enabled by default]
	  MUX_F(FIN_PLL, "fin_pll", fin_pll_p, OM_STAT, 0, 1,
	  ^
	drivers/clk/samsung/clk-s5pv210.c:412:2: warning: (near initialization for 'early_mux_clks[0].parent_names') [enabled by default]

That's because early_mux_clks[0].parent_names is of type const char **
while fin_pll_p as suggested doesn't provide an array of pointers to the
start of the contained strings.
I don't see a way to fix that unless we fix the maximal clock name
length globally and so waste much memory.

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-K?nig            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |

  reply	other threads:[~2014-09-12  7:59 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-09-11 21:04 [PATCH] clk: don't use __initconst for non-const arrays Uwe Kleine-König
2014-09-11 21:48 ` Tomasz Figa
2014-09-11 22:04 ` Uwe Kleine-König
2014-09-12  7:42   ` Ard Biesheuvel
2014-09-12  7:59     ` Uwe Kleine-König [this message]
  -- strict thread matches above, loose matches on Subject: below --
2015-02-18  9:59 Uwe Kleine-König
2015-03-25 10:41 ` Uwe Kleine-König
2015-04-13  0:20   ` Michael Turquette

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=20140912075952.GR3755@pengutronix.de \
    --to=u.kleine-koenig@pengutronix.de \
    --cc=linux-arm-kernel@lists.infradead.org \
    /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;
as well as URLs for NNTP newsgroup(s).