From: Paolo Bonzini <pbonzini@redhat.com>
To: Sergey Fedorov <serge.fdrv@gmail.com>,
sergey.fedorov@linaro.org, qemu-devel@nongnu.org
Cc: "Richard Henderson" <rth@twiddle.net>,
"Peter Crosthwaite" <crosthwaite.peter@gmail.com>,
"Alex Bennée" <alex.bennee@linaro.org>
Subject: Re: [Qemu-devel] [PATCH 4/5] tcg: reorder removal from lists in tb_phys_invalidate
Date: Thu, 14 Apr 2016 17:13:26 +0200 [thread overview]
Message-ID: <570FB396.6040703@redhat.com> (raw)
In-Reply-To: <570FACF1.6020009@gmail.com>
On 14/04/2016 16:45, Sergey Fedorov wrote:
> So what would you suggest to use for x86? I can't think of something
> that looks like a really compelling combination when I look at
> cpu_get_tb_cpu_state() in target-i386/cpu.h.
On x86 I think we should define HF_INVALID_TB to an invalid flag
combination. I can think of several solutions:
- #defining HF_INVALID_TB to an invalid combination (e.g. HF_CS64_MASK,
because it always appears together with HF_LMA_MASK; see code that
updates those hflags in cpu_x86_load_seg_cache, cpu_x86_update_cr0,
kvm_get_sregs). Advantage: doesn't waste a bit, reasonably self
documenting. Disadvantage: a bit tricky, but still my favorite.
- rename HF_SOFTMMU_MASK to HF_INVALID_MASK (it's always the same as
CONFIG_SOFTMMU so we can remove it), then #define HF_INVALID_TB
HF_INVALID_MASK. Advantage: obviously correct. Disadvantage: wastes a
bit. My second favorite.
- #defining HF_INVALID_TB to -1. Advantage: ?!? Disadvantage:
everything. Looks tame, actually a huge hack
- #defining HF_INVALID_TB to the "wrong" direction HF_SOFTMMU_MASK (i.e.
to 0 if CONFIG_SOFTMMU, . Advantage: obviously correct. Disadvantage:
huge hack, HF_SOFTMMU_MASK is unused anyway.
Choose your own favorite. :)
(Setting cs_base to -1 actually would work on 64-bit x86, but not on
qemu-system-i386).
> Personally, I'm not so
> happy trying to use pc/cs_base/flags to mark an invalid TB. Are my
> worries unreasonable? :)
Can you explain your worries?
The advantages are that it's O(1) and it obviously doesn't affect other
TBs than the invalidated one.
> Anyway, I am wondering if there is still a way to clear tb_phys_hash and
> tb_jmp_cache safely.
>
> Maybe something like this:
> * Remove the TB from physical hash list
So at this point tb_find_slow cannot find it.
> * Memory barrier
> * Remove the TB from each vCPU's virtual address hash cache
tb_find_fast then cannot find it either.
> Would that work?
This is very similar to the current code. From 10,000 feet, because
tb_find_fast calls tb_find_slow, this could indeed work, but I'm a bit
concerned about how to order the removal of the jump lists. The usage
of "tcg_ctx.tb_ctx.tb_invalidated_flag = 1" in the existing code was
what worries me. Indeed the motivation of this patch was removing that
single line of code to prepare for the move of tb_invalidated_flag to
CPUState.
Also, this loop will not be thread-safe anymore as soon as Fred's
"tb_jmp_cache lookup outside tb_lock" goes in:
CPU_FOREACH(cpu) {
if (cpu->tb_jmp_cache[h] == tb) {
cpu->tb_jmp_cache[h] = NULL;
}
}
It should use atomic_cmpxchg (slow!) or to unconditionally NULL out
cpu->tb_jmp_cache (a bit hacky). Preparing for that change is an added
bonus of the tb-hacking approach.
Paolo
next prev parent reply other threads:[~2016-04-14 15:13 UTC|newest]
Thread overview: 41+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-03-17 13:46 [Qemu-devel] [PATCH 0/5] tcg: Misc clean-up patches from Paolo sergey.fedorov
2016-03-17 13:46 ` [Qemu-devel] [PATCH 1/5] tcg: code_bitmap is not used by user-mode emulation sergey.fedorov
2016-03-17 14:56 ` Peter Maydell
2016-03-17 15:03 ` Sergey Fedorov
2016-03-17 13:46 ` [Qemu-devel] [PATCH 2/5] tcg: reorganize tb_find_physical loop sergey.fedorov
2016-03-17 14:59 ` Peter Maydell
2016-03-22 14:59 ` Alex Bennée
2016-03-22 15:00 ` Paolo Bonzini
2016-03-29 13:19 ` Sergey Fedorov
2016-03-29 13:26 ` Paolo Bonzini
2016-03-29 14:05 ` Sergey Fedorov
2016-03-29 14:26 ` Alex Bennée
2016-03-29 14:37 ` Sergey Fedorov
2016-03-17 13:46 ` [Qemu-devel] [PATCH 3/5] tcg: always keep jump target and tb->jmp_next consistent sergey.fedorov
2016-03-17 17:57 ` Richard Henderson
2016-03-17 19:31 ` Paolo Bonzini
2016-03-17 20:45 ` Sergey Fedorov
2016-03-17 20:46 ` Richard Henderson
2016-03-18 10:29 ` Sergey Fedorov
2016-03-18 10:32 ` Sergey Fedorov
2016-03-17 13:46 ` [Qemu-devel] [PATCH 4/5] tcg: reorder removal from lists in tb_phys_invalidate sergey.fedorov
2016-03-17 15:09 ` Paolo Bonzini
2016-03-17 15:14 ` Sergey Fedorov
2016-03-28 15:18 ` Sergey Fedorov
2016-03-28 21:21 ` Paolo Bonzini
2016-03-29 10:03 ` Sergey Fedorov
2016-03-29 10:37 ` Paolo Bonzini
2016-03-29 12:31 ` Sergey Fedorov
2016-03-29 13:43 ` Alex Bennée
2016-04-14 14:45 ` Sergey Fedorov
2016-04-14 15:13 ` Paolo Bonzini [this message]
2016-04-14 15:36 ` Sergey Fedorov
2016-04-14 17:27 ` Paolo Bonzini
2016-04-14 18:29 ` Sergey Fedorov
2016-04-14 18:37 ` Sergey Fedorov
2016-03-28 18:42 ` Sergey Fedorov
2016-03-28 20:58 ` Paolo Bonzini
2016-03-29 0:17 ` Richard Henderson
2016-03-17 13:46 ` [Qemu-devel] [PATCH 5/5] tcg: move tb_invalidated_flag to CPUState sergey.fedorov
2016-03-22 15:07 ` Alex Bennée
2016-03-22 15:11 ` Sergey Fedorov
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=570FB396.6040703@redhat.com \
--to=pbonzini@redhat.com \
--cc=alex.bennee@linaro.org \
--cc=crosthwaite.peter@gmail.com \
--cc=qemu-devel@nongnu.org \
--cc=rth@twiddle.net \
--cc=serge.fdrv@gmail.com \
--cc=sergey.fedorov@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).