Mailing-List: contact dev-help@httpd.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@httpd.apache.org
Content-Type: text/plain; charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 8.2 \(2104\))
Subject: Re: svn commit: r1706669 - in /httpd/httpd/trunk: ./ include/
 modules/http/ modules/ssl/ server/ server/mpm/event/ server/mpm/motorz/
 server/mpm/simple/
From: Jim Jagielski <jim@jaguNET.com>
In-Reply-To: <07809A63-C0DB-44D2-86AE-1505E5F46B97@sharp.fm>
Date: Tue, 6 Oct 2015 13:56:31 -0400
Content-Transfer-Encoding: quoted-printable
Message-Id: <1AC4D4CB-D6D5-4517-BED3-E922526CC6E6@jaguNET.com>
References: <20151004101052.A6E193A0234@svn01-us-west.apache.org>
 <CAKQ1sVNRVJqqZSmzVn86K3SbztGSBQ68D2kTJRhWU1qMWDrUYA@mail.gmail.com>
 <9A9FBD0C-B503-4CDD-A833-74C69FF248EA@sharp.fm>
 <CAKQ1sVN2WwBhbJ26JOdm7YFnMyXtOfAbCX1eCeUse7Zh1k8O5Q@mail.gmail.com>
 <CAKQ1sVOA-X7CpjeJruq7wgbFi+WaBsMn-X_0Bpw+CUxmnHtYtg@mail.gmail.com>
 <660C359D-89D0-4951-ACEA-879BA3F4D726@sharp.fm>
 <CAKQ1sVNCADr1ojLpZ3R19q2tuyXqMGXO8OP-RWv07uQ+4qZsSQ@mail.gmail.com>
 <CAKQ1sVPVuYQEynjevqQyP-AFe3GFuT10-dvxY+jvWneouOYC-g@mail.gmail.com>
 <CAKQ1sVO5J3SBqc8w=m16R=vESgTpKqL5_E7dD9TRXMdd1KbT8A@mail.gmail.com>
 <07809A63-C0DB-44D2-86AE-1505E5F46B97@sharp.fm>
To: dev@httpd.apache.org


> On Oct 6, 2015, at 1:29 PM, Graham Leggett <minfrin@sharp.fm> wrote:
>=20
> On 06 Oct 2015, at 7:00 PM, Yann Ylavic <ylavic.dev@gmail.com> wrote:
>=20
>> Yet another filter which does not remove itself after the EOS, and
>> destroys the EOR bucket while still using *r after.
>=20
> I am wondering if the EOR bucket is a suboptimal way to clean up after =
ourselves.
>=20
> The MPM itself is in a position to know when the filter buffers are =
all empty and it is safe to delete a request pool. I am imagining a =
=E2=80=9Cshutdown set=E2=80=9D, which contains the pools which have been =
marked as ready to shut down. If there is a pool in the shutdown set, =
walk the outstanding filters and look for any filters whose pools are =
descendants of that pool, and check if the filters are empty. If so, =
apr_pool_destroy() and we=E2=80=99re done. If the =E2=80=9Cshutdown =
set=E2=80=9D is empty, do nothing.
>=20
> Instead of creating an EOR bucket, we call an MPM API that says =E2=80=9C=
add this pool to the shutdown set=E2=80=9D, and then forget about it.
>=20
> Legacy MPMs fall back to EOR behaviour, but we deprecate it.
>=20
> I think the above is a far safer approach than assuming filters will =
=E2=80=9Cdo the right thing=E2=80=9D.
>=20

Basic idea seems sound... +1