From: Paolo Bonzini <pbonzini@redhat.com>
To: Richard Henderson <richard.henderson@linaro.org>, qemu-devel@nongnu.org
Cc: peter.maydell@linaro.org
Subject: Re: [PATCH 0/7] exec: Improve code for TARGET_PAGE_BITS_VARY
Date: Fri, 20 Sep 2019 09:30:18 +0200 [thread overview]
Message-ID: <6e6c8fce-3805-8fe2-5505-92c385e7103b@redhat.com> (raw)
In-Reply-To: <20190919232952.6382-1-richard.henderson@linaro.org>
On 20/09/19 01:29, Richard Henderson wrote:
> There's currently a fair amount of overhead in the way we currently
> treat TARGET_PAGE_{BITS,SIZE,MASK} with TARGET_PAGE_BITS_VARY.
>
> We have assertions that TARGET_PAGE_BITS has been finalized. Which
> is fine, but the variable that controls the assertion may be assumed
> to be modified by any function call, which means that we have lots
> of duplicate assertions.
>
> This re-arranges things using a const symbol, which allows the compiler
> to assume that the variable is not modified across calls. In order to
> allow initialization of the variable during startup, use an alias that
> is non-const and controls the allocation into a read-write section.
>
> Remove the assertion for release builds.
>
> Precompute TARGET_PAGE_MASK. This removes a runtime shift and allows
> the variable to be used as a direct memory operand on x86.
>
> Size reductions vs master for qemu-system-aarch64 for various hosts:
>
> PPC64LE:
> debug-tcg: -32264
> release: -44360
> AARCH64:
> debug-tcg: -33304
> relase: -77080
> X86_64:
> debug-tcg: -6685
> relase: -15597
>
>
> r~
>
>
> Richard Henderson (7):
> exec: Use TARGET_PAGE_BITS_MIN for TLB flags
> exec: Split out variable page size support to exec-vary.c
> exec: Use const alias for TARGET_PAGE_BITS_VARY
> exec: Restrict TARGET_PAGE_BITS_VARY assert to CONFIG_DEBUG_TCG
> exec: Promote TARGET_PAGE_MASK to target_long
> exec: Tidy TARGET_PAGE_ALIGN
> exec: Cache TARGET_PAGE_MASK for TARGET_PAGE_BITS_VARY
>
> Makefile.target | 2 +-
> include/exec/cpu-all.h | 41 +++++++++++++-------
> include/qemu-common.h | 6 +++
> exec-vary.c | 88 ++++++++++++++++++++++++++++++++++++++++++
> exec.c | 34 ----------------
> 5 files changed, 123 insertions(+), 48 deletions(-)
> create mode 100644 exec-vary.c
>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
next prev parent reply other threads:[~2019-09-20 7:32 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-19 23:29 [PATCH 0/7] exec: Improve code for TARGET_PAGE_BITS_VARY Richard Henderson
2019-09-19 23:29 ` [PATCH 1/7] exec: Use TARGET_PAGE_BITS_MIN for TLB flags Richard Henderson
2019-09-19 23:29 ` [PATCH 2/7] exec: Split out variable page size support to exec-vary.c Richard Henderson
2019-09-21 8:59 ` Philippe Mathieu-Daudé
2019-09-19 23:29 ` [PATCH 3/7] exec: Use const alias for TARGET_PAGE_BITS_VARY Richard Henderson
2019-09-19 23:29 ` [PATCH 4/7] exec: Restrict TARGET_PAGE_BITS_VARY assert to CONFIG_DEBUG_TCG Richard Henderson
2019-09-21 9:01 ` Philippe Mathieu-Daudé
2019-09-19 23:29 ` [PATCH 5/7] exec: Promote TARGET_PAGE_MASK to target_long Richard Henderson
2019-09-19 23:29 ` [PATCH 6/7] exec: Tidy TARGET_PAGE_ALIGN Richard Henderson
2019-09-19 23:29 ` [PATCH 7/7] exec: Cache TARGET_PAGE_MASK for TARGET_PAGE_BITS_VARY Richard Henderson
2019-09-20 7:30 ` Paolo Bonzini [this message]
2019-09-21 0:32 ` [PATCH 0/7] exec: Improve code " no-reply
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=6e6c8fce-3805-8fe2-5505-92c385e7103b@redhat.com \
--to=pbonzini@redhat.com \
--cc=peter.maydell@linaro.org \
--cc=qemu-devel@nongnu.org \
--cc=richard.henderson@linaro.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).