From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 88171 invoked by alias); 11 Dec 2019 15:41:21 -0000 Mailing-List: contact dwz-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Post: List-Help: List-Subscribe: Sender: dwz-owner@sourceware.org Received: (qmail 88160 invoked by uid 89); 11 Dec 2019 15:41:21 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Checked: by ClamAV 0.100.3 on sourceware.org X-Virus-Found: No X-Spam-SWARE-Status: No, score=-24.2 required=5.0 tests=AWL,BAYES_00,GIT_PATCH_0,GIT_PATCH_1,GIT_PATCH_2,GIT_PATCH_3,LIKELY_SPAM_BODY,SPF_PASS autolearn=ham version=3.3.1 spammy=ack, HX-Languages-Length:3216, million X-Spam-Status: No, score=-24.2 required=5.0 tests=AWL,BAYES_00,GIT_PATCH_0,GIT_PATCH_1,GIT_PATCH_2,GIT_PATCH_3,LIKELY_SPAM_BODY,SPF_PASS autolearn=ham version=3.3.1 X-Spam-Checker-Version: SpamAssassin 3.3.1 (2010-03-16) on sourceware.org X-Spam-Level: X-HELO: mx1.suse.de X-Virus-Scanned: by amavisd-new at test-mx.suse.de Subject: Re: [RFC 1/13][odr] Cover letter To: Michael Matz Cc: dwz@sourceware.org, jakub@redhat.com References: <20191210171830.GA13804@delia> From: Tom de Vries Autocrypt: addr=tdevries@suse.de; keydata= xsBNBF0ltCcBCADDhsUnMMdEXiHFfqJdXeRvgqSEUxLCy/pHek88ALuFnPTICTwkf4g7uSR7 HvOFUoUyu8oP5mNb4VZHy3Xy8KRZGaQuaOHNhZAT1xaVo6kxjswUi3vYgGJhFMiLuIHdApoc u5f7UbV+egYVxmkvVLSqsVD4pUgHeSoAcIlm3blZ1sDKviJCwaHxDQkVmSsGXImaAU+ViJ5l CwkvyiiIifWD2SoOuFexZyZ7RUddLosgsO0npVUYbl6dEMq2a5ijGF6/rBs1m3nAoIgpXk6P TCKlSWVW6OCneTaKM5C387972qREtiArTakRQIpvDJuiR2soGfdeJ6igGA1FZjU+IsM5ABEB AAHNH1RvbSBkZSBWcmllcyA8dGRldnJpZXNAc3VzZS5kZT7CwKsEEwEIAD4WIQSsnSe5hKbL MK1mGmjuhV2rbOJEoAUCXSW0JwIbAwUJA8JnAAULCQgHAgYVCgkICwIEFgIDAQIeAQIXgAAh CRDuhV2rbOJEoBYhBKydJ7mEpsswrWYaaO6FXats4kSgc48H/Ra2lq5p3dHsrlQLqM7N68Fo eRDf3PMevXyMlrCYDGLVncQwMw3O/AkousktXKQ42DPJh65zoXB22yUt8m0g12xkLax98KFJ 5NyUloa6HflLl+wQL/uZjIdNUQaHQLw3HKwRMVi4l0/Jh/TygYG1Dtm8I4o708JS4y8GQxoQ UL0z1OM9hyM3gI2WVTTyprsBHy2EjMOu/2Xpod95pF8f90zBLajy6qXEnxlcsqreMaqmkzKn 3KTZpWRxNAS/IH3FbGQ+3RpWkNGSJpwfEMVCeyK5a1n7yt1podd1ajY5mA1jcaUmGppqx827 8TqyteNe1B/pbiUt2L/WhnTgW1NC1QDOwE0EXSW0JwEIAM99H34Bu4MKM7HDJVt864MXbx7B 1M93wVlpJ7Uq+XDFD0A0hIal028j+h6jA6bhzWto4RUfDl/9mn1StngNVFovvwtfzbamp6+W pKHZm9X5YvlIwCx131kTxCNDcF+/adRW4n8CU3pZWYmNVqhMUiPLxElA6QhXTtVBh1RkjCZQ Kmbd1szvcOfaD8s+tJABJzNZsmO2hVuFwkDrRN8Jgrh92a+yHQPd9+RybW2l7sJv26nkUH5Z 5s84P6894ebgimcprJdAkjJTgprl1nhgvptU5M9Uv85Pferoh2groQEAtRPlCGrZ2/2qVNe9 XJfSYbiyedvApWcJs5DOByTaKkcAEQEAAcLAkwQYAQgAJhYhBKydJ7mEpsswrWYaaO6FXats 4kSgBQJdJbQnAhsMBQkDwmcAACEJEO6FXats4kSgFiEErJ0nuYSmyzCtZhpo7oVdq2ziRKD3 twf7BAQBZ8TqR812zKAD7biOnWIJ0McV72PFBxmLIHp24UVe0ZogtYMxSWKLg3csh0yLVwc7 H3vldzJ9AoK3Qxp0Q6K/rDOeUy3HMqewQGcqrsRRh0NXDIQk5CgSrZslPe47qIbe3O7ik/MC q31FNIAQJPmKXX25B115MMzkSKlv4udfx7KdyxHrTSkwWZArLQiEZj5KG4cCKhIoMygPTA3U yGaIvI/BGOtHZ7bEBVUCFDFfOWJ26IOCoPnSVUvKPEOH9dv+sNy7jyBsP5QxeTqwxC/1ZtNS DUCSFQjqA6bEGwM22dP8OUY6SC94x1G81A9/xbtm9LQxKm0EiDH8KBMLfQ== Message-ID: <44919257-1313-ed65-fb88-43b377a9bb12@suse.de> Date: Tue, 01 Jan 2019 00:00:00 -0000 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:68.0) Gecko/20100101 Thunderbird/68.2.1 MIME-Version: 1.0 In-Reply-To: Content-Type: multipart/mixed; boundary="------------36EBBE3A6707BF68CD9BA3C1" Content-Language: en-US X-SW-Source: 2019-q4/txt/msg00148.txt.bz2 This is a multi-part message in MIME format. --------------36EBBE3A6707BF68CD9BA3C1 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: 7bit Content-length: 1707 On 11-12-2019 14:15, Michael Matz wrote: > Hello, > > On Tue, 10 Dec 2019, Tom de Vries wrote: > >>> Seriously, this is nice. The additional reduction of using the ODR rule >>> on cc1 isn't as big as I thought in the past, probably the largest stuff >>> in .debug_info aren't type descriptions anymore, but still, it's a 22% >>> reduction on top, so nothing to sneeze at. (Maybe for firefox it's more, >>> I remember Honza moaning about many multiple struct type chains that >>> only differed in the completeness of their pointer members) >> >> I'd hope that the --odr-mode=unify in a future patch series will bring >> further improvement, but we'll have to see how much. >> >> As for this patch series, I'm planning to commit in a couple of days, >> unless there are serious objections. >> >> Could you at least review the user visible parts? Command line option >> names, man page entries (and usage ... oops, I forgot to update usage). > That's done: https://sourceware.org/ml/dwz/2019-q4/msg00147.html . > If you mean me: I'm fine with the names, Ack. > but would expand a little bit on > the explanations. For --odr I would add a sentence describing the effect > of applying the ODR rule (without going into too much detail), i.e. > mention that this causes type DIEs that have to be the same per C++ rules > (same name basically) to be more forcefully deduped. Also a > description of the two modes basic/link might be in order for the user to > make a meaningful choice; alteratively describe it as debug/development > option if the user isn't supposed to use it in regular use. > I've updated the man page entries with your suggestions. Better like this? Thanks, - Tom --------------36EBBE3A6707BF68CD9BA3C1 Content-Type: text/x-patch; charset=UTF-8; name="0013-odr-Add-odr-no-odr-and-odr-mode-entries-to-man-page.patch" Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename*0="0013-odr-Add-odr-no-odr-and-odr-mode-entries-to-man-page.pat"; filename*1="ch" Content-length: 1511 [odr] Add --odr/--no-odr and --odr-mode entries to man page Add an --odr/--no-odr entry and a --odr-mode entry to the man page, advertising the optimization to the user. 2019-12-10 Tom de Vries * dwz.1: Add --odr/--no-odr and a --odr-mode entries. --- dwz.1 | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/dwz.1 b/dwz.1 index a5a1ef6..3df7e59 100644 --- a/dwz.1 +++ b/dwz.1 @@ -88,6 +88,23 @@ containing more than \fICOUNT\fR DIEs at all. The default is 50 million DIEs. Specifying none as argument disables the limit. .TP +.B \-\-odr / \-\-no-odr +Enable/disable One-Definition-Rule optimization for C++ compilation units. +This optimization causes struct/union/class DIEs with the same name to be +considered equal. This has the effect that DIEs referring to distinct DIEs +representing the same type (like f.i. pointer type DIEs) are considered equal, +and may be deduplicated. +Enabled by default. +.TP +.B \-\-odr-mode= +Set the One-Definition-Rule optimization aggressiveness: basic or link. +When using the link setting, the optimization will attempt to replace +declarations of a struct/union/class with a corresponding definition. When +using the basic setting, that part of the optimization is disabled. +In normal operation, the link setting should be used. The basic setting is +provided only as fallback in case of problems with the link setting. Set to +link by default. +.TP .B \-? \-\-help Print short help and exit. .TP --------------36EBBE3A6707BF68CD9BA3C1--