From: "Alex Bennée" <alex.bennee@linaro.org>
To: Richard Henderson <richard.henderson@linaro.org>
Cc: qemu-devel@nongnu.org
Subject: Re: [PATCH v3 18/28] tcg: Tidy tcg_n_regions
Date: Tue, 08 Jun 2021 17:06:32 +0100 [thread overview]
Message-ID: <87v96ojq78.fsf@linaro.org> (raw)
In-Reply-To: <20210502231844.1977630-19-richard.henderson@linaro.org>
Richard Henderson <richard.henderson@linaro.org> writes:
> Compute the value using straight division and bounds,
> rather than a loop. Pass in tb_size rather than reading
> from tcg_init_ctx.code_gen_buffer_size,
>
> Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
> ---
> tcg/region.c | 29 ++++++++++++-----------------
> 1 file changed, 12 insertions(+), 17 deletions(-)
>
> diff --git a/tcg/region.c b/tcg/region.c
> index bd81b35359..b44246e1aa 100644
> --- a/tcg/region.c
> +++ b/tcg/region.c
> @@ -363,38 +363,33 @@ void tcg_region_reset_all(void)
> tcg_region_tree_reset_all();
> }
>
> -static size_t tcg_n_regions(unsigned max_cpus)
> +static size_t tcg_n_regions(size_t tb_size, unsigned max_cpus)
> {
> #ifdef CONFIG_USER_ONLY
> return 1;
> #else
> + size_t n_regions;
> +
> /*
> * It is likely that some vCPUs will translate more code than others,
> * so we first try to set more regions than max_cpus, with those regions
> * being of reasonable size. If that's not possible we make do by evenly
> * dividing the code_gen_buffer among the vCPUs.
> */
> - size_t i;
> -
> /* Use a single region if all we have is one vCPU thread */
> if (max_cpus == 1 || !qemu_tcg_mttcg_enabled()) {
> return 1;
> }
>
> - /* Try to have more regions than max_cpus, with each region being >= 2 MB */
> - for (i = 8; i > 0; i--) {
> - size_t regions_per_thread = i;
> - size_t region_size;
> -
> - region_size = tcg_init_ctx.code_gen_buffer_size;
> - region_size /= max_cpus * regions_per_thread;
> -
> - if (region_size >= 2 * 1024u * 1024) {
> - return max_cpus * regions_per_thread;
> - }
> + /*
> + * Try to have more regions than max_cpus, with each region being >= 2 MB.
> + * If we can't, then just allocate one region per vCPU thread.
> + */
> + n_regions = tb_size / (2 * MiB);
> + if (n_regions <= max_cpus) {
> + return max_cpus;
> }
> - /* If we can't, then just allocate one region per vCPU thread */
> - return max_cpus;
> + return MIN(n_regions, max_cpus * 8);
> #endif
> }
This is so much easier to follow now ;-)
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
--
Alex Bennée
next prev parent reply other threads:[~2021-06-08 16:09 UTC|newest]
Thread overview: 92+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-05-02 23:18 [PATCH v3 00/28] tcg: Clean up code_gen_buffer allocation Richard Henderson
2021-05-02 23:18 ` [PATCH v3 01/28] meson: Split out tcg/meson.build Richard Henderson
2021-06-07 19:38 ` Alex Bennée
2021-06-09 14:56 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 02/28] meson: Split out fpu/meson.build Richard Henderson
2021-06-07 19:38 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 03/28] tcg: Re-order tcg_region_init vs tcg_prologue_init Richard Henderson
2021-06-08 9:04 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 04/28] tcg: Remove error return from tcg_region_initial_alloc__locked Richard Henderson
2021-06-08 11:04 ` Alex Bennée
2021-06-08 15:44 ` Richard Henderson
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 05/28] tcg: Split out tcg_region_initial_alloc Richard Henderson
2021-06-08 11:45 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 06/28] tcg: Split out tcg_region_prologue_set Richard Henderson
2021-06-08 11:47 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 07/28] tcg: Split out region.c Richard Henderson
2021-06-08 11:48 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 08/28] accel/tcg: Inline cpu_gen_init Richard Henderson
2021-06-08 11:49 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 09/28] accel/tcg: Move alloc_code_gen_buffer to tcg/region.c Richard Henderson
2021-06-08 11:50 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 10/28] accel/tcg: Rename tcg_init to tcg_init_machine Richard Henderson
2021-06-08 11:51 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 11/28] tcg: Create tcg_init Richard Henderson
2021-06-08 11:52 ` Alex Bennée
2021-06-09 14:57 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 12/28] accel/tcg: Merge tcg_exec_init into tcg_init_machine Richard Henderson
2021-06-08 11:55 ` Alex Bennée
2021-06-08 15:45 ` Richard Henderson
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-06-10 15:16 ` Richard Henderson
2021-05-02 23:18 ` [PATCH v3 13/28] accel/tcg: Pass down max_cpus to tcg_init Richard Henderson
2021-06-08 15:44 ` Alex Bennée
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 14/28] tcg: Introduce tcg_max_ctxs Richard Henderson
2021-06-08 15:54 ` Alex Bennée
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 15/28] tcg: Move MAX_CODE_GEN_BUFFER_SIZE to tcg-target.h Richard Henderson
2021-06-08 15:56 ` Alex Bennée
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 16/28] tcg: Replace region.end with region.total_size Richard Henderson
2021-06-08 16:03 ` Alex Bennée
2021-06-08 16:12 ` Richard Henderson
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 17/28] tcg: Rename region.start to region.after_prologue Richard Henderson
2021-06-08 16:06 ` Alex Bennée
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 18/28] tcg: Tidy tcg_n_regions Richard Henderson
2021-06-08 16:06 ` Alex Bennée [this message]
2021-06-09 14:58 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 19/28] tcg: Tidy split_cross_256mb Richard Henderson
2021-06-09 10:01 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-06-10 15:20 ` Richard Henderson
2021-05-02 23:18 ` [PATCH v3 20/28] tcg: Move in_code_gen_buffer and tests to region.c Richard Henderson
2021-06-09 10:07 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 21/28] tcg: Allocate code_gen_buffer into struct tcg_region_state Richard Henderson
2021-06-09 10:29 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 22/28] tcg: Return the map protection from alloc_code_gen_buffer Richard Henderson
2021-06-09 10:30 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 23/28] tcg: Sink qemu_madvise call to common code Richard Henderson
2021-06-09 10:34 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 24/28] util/osdep: Add qemu_mprotect_rw Richard Henderson
2021-06-09 10:34 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 25/28] tcg: Round the tb_size default from qemu_get_host_physmem Richard Henderson
2021-06-09 10:36 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 26/28] tcg: Merge buffer protection and guard page protection Richard Henderson
2021-06-09 10:39 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 27/28] tcg: When allocating for !splitwx, begin with PROT_NONE Richard Henderson
2021-06-09 11:21 ` Alex Bennée
2021-06-10 15:34 ` Richard Henderson
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:18 ` [PATCH v3 28/28] tcg: Move tcg_init_ctx and tcg_ctx from accel/tcg/ Richard Henderson
2021-06-09 11:30 ` Alex Bennée
2021-06-09 14:59 ` Luis Fernando Fujita Pires
2021-05-02 23:45 ` [PATCH v3 00/28] tcg: Clean up code_gen_buffer allocation 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=87v96ojq78.fsf@linaro.org \
--to=alex.bennee@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 read-only IMAP folder(s) and NNTP newsgroup(s).