From: Ferry Huberts <mailings@hupie.com>
To: Avi Kivity <avi@redhat.com>
Cc: Jan Kiszka <jan.kiszka@siemens.com>,
qemu-devel@nongnu.org, kvm@vger.kernel.org
Subject: Re: [PATCH] memory: transaction API
Date: Thu, 21 Jul 2011 14:26:03 +0200 [thread overview]
Message-ID: <4E281ADB.90708@hupie.com> (raw)
In-Reply-To: <4E281684.5000203@redhat.com>
On 07/21/2011 02:07 PM, Avi Kivity wrote:
> On 07/21/2011 02:04 PM, Ferry Huberts wrote:
>> On 07/21/2011 12:21 PM, Avi Kivity wrote:
>> > Allow changes to the memory hierarchy to be accumulated and
>> > made visible all at once. This reduces computational effort,
>> > especially when an accelerator (e.g. kvm) is involved.
>> >
>> > Useful when a single register update causes multiple changes
>> > to an address space.
>> >
>> >
>> > +void memory_region_transaction_begin(void)
>> > +{
>> > + ++memory_region_transaction_depth;
>> > +}
>> > +
>>
>> wouldn't you rather keep it safe by doing either here
>>
>> if (!memory_region_transaction_depth)
>> memory_region_transaction_depth++;
>>
>
> Why? I want to allow nesting transactions (not that I anticipate such a
> case).
>
>> > +void memory_region_transaction_commit(void)
>> > +{
>> > + if (!memory_region_transaction_depth) {
>> > + abort();
>> > + }
>>
>>
>> > + --memory_region_transaction_depth;
>> > + memory_region_update_topology();
>> > +}
>> > +
>>
>> or by doing here
>>
>> while (!memory_region_transaction_depth)
>> memory_region_transaction_depth--;
doesn't memory_region_update_topology commit all accumulated changes? if
it does then memory_region_transaction_depth is left non-zero in the
nesting case while no more changes are actually present, resulting in
superfluous calls to memory_region_update_topology.
maybe I misunderstood memory_region_update_topology?
>>
>>
>> with your setup nesting transactions will not work correctly I think.
>> You seem to have designed it to not do nesting, so it's safer to make
>> that explicit in your code imho
>
> Nesting should work just fine.
>
>> therefore I'd go for the change in _begin
>>
>> also, wouldn't you rather rename memory_region_transaction_depth to
>> memory_region_transaction_pending?
>
> The existing name works for me, but if people want it changed, that's
> fine too.
>
>
--
Ferry Huberts
next prev parent reply other threads:[~2011-07-21 12:26 UTC|newest]
Thread overview: 21+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-07-21 10:21 [PATCH] memory: transaction API Avi Kivity
2011-07-21 10:38 ` Jan Kiszka
2011-07-21 12:05 ` Avi Kivity
2011-07-21 12:08 ` Avi Kivity
2011-07-21 12:09 ` Jan Kiszka
2011-07-21 12:13 ` Avi Kivity
2011-07-21 12:52 ` Jan Kiszka
2011-07-21 12:58 ` Avi Kivity
2011-07-21 13:17 ` Jan Kiszka
2011-07-21 13:50 ` Avi Kivity
2011-07-21 14:32 ` Jan Kiszka
2011-07-21 14:39 ` Avi Kivity
2011-07-21 15:05 ` Jan Kiszka
2011-07-21 15:05 ` [Qemu-devel] " Jan Kiszka
2011-07-21 15:11 ` Avi Kivity
2011-07-21 15:11 ` [Qemu-devel] " Avi Kivity
2011-07-21 11:04 ` Ferry Huberts
2011-07-21 12:07 ` Avi Kivity
2011-07-21 12:26 ` Ferry Huberts [this message]
2011-07-21 12:46 ` Avi Kivity
2011-07-21 12:56 ` Ferry Huberts
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=4E281ADB.90708@hupie.com \
--to=mailings@hupie.com \
--cc=avi@redhat.com \
--cc=jan.kiszka@siemens.com \
--cc=kvm@vger.kernel.org \
--cc=qemu-devel@nongnu.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.