All of lore.kernel.org
 help / color / mirror / Atom feed
From: Paul Burton <paul.burton@mips.com>
To: Hauke Mehrtens <hauke@hauke-m.de>
Cc: ralf@linux-mips.org, jhogan@kernel.org,
	linux-mips@linux-mips.org, ak@linux.intel.com
Subject: Re: [PATCH] MIPS: remove const from mips_io_port_base
Date: Thu, 28 Jun 2018 15:48:08 -0700	[thread overview]
Message-ID: <20180628224808.osbostnj77pzrsod@pburton-laptop> (raw)
In-Reply-To: <20180616154745.28230-1-hauke@hauke-m.de>

Hi Hauke,

On Sat, Jun 16, 2018 at 05:47:45PM +0200, Hauke Mehrtens wrote:
> This variable is changed by some platform init code. When LTO is used
> gcc assumes that the variable never changes and inlines the constant
> instead of checking this variable which is wrong.
> 
> This fixes a runtime boot problem when LTO is activated.
> 
> Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
> ---
>  arch/mips/include/asm/io.h | 2 +-
>  arch/mips/kernel/setup.c   | 2 +-
>  2 files changed, 2 insertions(+), 2 deletions(-)

This does cause GCC 8.1.0 to emit extra loads, adding 652 bytes to a
32r2el_defconfig build from current mips-next. I'd expect kernels for
some of the older supported machines might show a larger gain as they're
more likely to use inX/outX. Combined with the fact that LTO support
isn't in mainline yet anyway, this makes me hesitant to apply this one.

Although it is tempting from the standpoint that set_io_port_base() is
an ugly hack & this would allow its removal... I wonder if we could
clean that up in the same patch though, since it's a natural consequence
of removing const & might make it attractive enough to tolerate the code
size increase.

Even better might be if we could use a fixed virtual address for the I/O
port base, for example by using fixmap. That should give even better
code generation but would of course have TLB overhead & be more
complicated to implement.

Thanks,
    Paul

> diff --git a/arch/mips/include/asm/io.h b/arch/mips/include/asm/io.h
> index a7d0b836f2f7..f28f8cd44dd3 100644
> --- a/arch/mips/include/asm/io.h
> +++ b/arch/mips/include/asm/io.h
> @@ -60,7 +60,7 @@
>   * instruction, so the lower 16 bits must be zero.  Should be true on
>   * on any sane architecture; generic code does not use this assumption.
>   */
> -extern const unsigned long mips_io_port_base;
> +extern unsigned long mips_io_port_base;
>  
>  /*
>   * Gcc will generate code to load the value of mips_io_port_base after each
> diff --git a/arch/mips/kernel/setup.c b/arch/mips/kernel/setup.c
> index 2c96c0c68116..153460c531a9 100644
> --- a/arch/mips/kernel/setup.c
> +++ b/arch/mips/kernel/setup.c
> @@ -75,7 +75,7 @@ static char __initdata builtin_cmdline[COMMAND_LINE_SIZE] = CONFIG_CMDLINE;
>   * mips_io_port_base is the begin of the address space to which x86 style
>   * I/O ports are mapped.
>   */
> -const unsigned long mips_io_port_base = -1;
> +unsigned long mips_io_port_base = -1;
>  EXPORT_SYMBOL(mips_io_port_base);
>  
>  static struct resource code_resource = { .name = "Kernel code", };
> -- 
> 2.11.0
> 

      reply	other threads:[~2018-06-28 22:48 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-06-16 15:47 [PATCH] MIPS: remove const from mips_io_port_base Hauke Mehrtens
2018-06-28 22:48 ` Paul Burton [this message]

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=20180628224808.osbostnj77pzrsod@pburton-laptop \
    --to=paul.burton@mips.com \
    --cc=ak@linux.intel.com \
    --cc=hauke@hauke-m.de \
    --cc=jhogan@kernel.org \
    --cc=linux-mips@linux-mips.org \
    --cc=ralf@linux-mips.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 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.