From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from userp2130.oracle.com (userp2130.oracle.com [156.151.31.86]) by sourceware.org (Postfix) with ESMTPS id 64910384A02B for ; Wed, 25 Nov 2020 17:41:17 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.3.2 sourceware.org 64910384A02B Received: from pps.filterd (userp2130.oracle.com [127.0.0.1]) by userp2130.oracle.com (8.16.0.42/8.16.0.42) with SMTP id 0APHUaCi116174; Wed, 25 Nov 2020 17:41:14 GMT Received: from userp3030.oracle.com (userp3030.oracle.com [156.151.31.80]) by userp2130.oracle.com with ESMTP id 351kwhjn86-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=FAIL); Wed, 25 Nov 2020 17:41:14 +0000 Received: from pps.filterd (userp3030.oracle.com [127.0.0.1]) by userp3030.oracle.com (8.16.0.42/8.16.0.42) with SMTP id 0APHV1bx020529; Wed, 25 Nov 2020 17:41:13 GMT Received: from aserv0121.oracle.com (aserv0121.oracle.com [141.146.126.235]) by userp3030.oracle.com with ESMTP id 351kwen1td-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=OK); Wed, 25 Nov 2020 17:41:13 +0000 Received: from abhmp0002.oracle.com (abhmp0002.oracle.com [141.146.116.8]) by aserv0121.oracle.com (8.14.4/8.13.8) with ESMTP id 0APHfCaR009282; Wed, 25 Nov 2020 17:41:12 GMT Received: from dhcp-10-154-123-81.vpn.oracle.com (/10.154.123.81) by default (Oracle Beehive Gateway v4.0) with ESMTP ; Wed, 25 Nov 2020 09:41:12 -0800 From: Qing Zhao Message-Id: Mime-Version: 1.0 (Mac OS X Mail 13.4 \(3608.120.23.2.4\)) Subject: Re: How to traverse all the local variables that declared in the current routine? Date: Wed, 25 Nov 2020 11:41:10 -0600 In-Reply-To: Cc: Richard Sandiford , gcc Patches To: Richard Biener References: <217BE64F-A623-4453-B45B-D38B66B71B72@ORACLE.COM> <15EA64C7-D75F-4CE1-92C8-6940186A512A@ORACLE.COM> X-Mailer: Apple Mail (2.3608.120.23.2.4) X-Proofpoint-Virus-Version: vendor=nai engine=6000 definitions=9816 signatures=668682 X-Proofpoint-Spam-Details: rule=notspam policy=default score=0 bulkscore=0 spamscore=0 mlxlogscore=999 adultscore=0 malwarescore=0 suspectscore=3 phishscore=0 mlxscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2009150000 definitions=main-2011250110 X-Proofpoint-Virus-Version: vendor=nai engine=6000 definitions=9816 signatures=668682 X-Proofpoint-Spam-Details: rule=notspam policy=default score=0 impostorscore=0 malwarescore=0 lowpriorityscore=0 adultscore=0 priorityscore=1501 suspectscore=3 bulkscore=0 spamscore=0 mlxlogscore=999 mlxscore=0 clxscore=1015 phishscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2009150000 definitions=main-2011250110 X-Spam-Status: No, score=-5.5 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, HTML_MESSAGE, RCVD_IN_MSPIKE_H2, SPF_HELO_PASS, SPF_PASS, TXREP, UNPARSEABLE_RELAY autolearn=ham autolearn_force=no version=3.4.2 X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on server2.sourceware.org Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-Content-Filtered-By: Mailman/MimeDel 2.1.29 X-BeenThere: gcc-patches@gcc.gnu.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gcc-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 25 Nov 2020 17:41:19 -0000 > On Nov 25, 2020, at 3:11 AM, Richard Biener = wrote: >>=20 >>=20 >> Hi, >>=20 >> Does gcc provide an iterator to traverse all the local variables that = are declared in the current routine? >>=20 >> If not, what=E2=80=99s the best way to traverse the local variables? >>=20 >>=20 >> Depends on what for. There's the source level view you get by = walking >> BLOCK_VARS of the >> scope tree, theres cfun->local_variables (FOR_EACH_LOCAL_DECL) and >> there's SSA names >> (FOR_EACH_SSA_NAME). >>=20 >>=20 >> I am planing to add a new phase immediately after = =E2=80=9Cpass_late_warn_uninitialized=E2=80=9D to initialize all = auto-variables that are >> not explicitly initialized in the declaration, the basic idea is = following: >>=20 >> ** The proposal: >>=20 >> A. add a new GCC option: (same name and meaning as CLANG) >> -ftrivial-auto-var-init=3D[pattern|zero], similar pattern init as = CLANG; >>=20 >> B. add a new attribute for variable: >> __attribute((uninitialized) >> the marked variable is uninitialized intentionaly for performance = purpose. >>=20 >> C. The implementation needs to keep the current static warning on = uninitialized >> variables untouched in order to avoid "forking the language". >>=20 >>=20 >> ** The implementation: >>=20 >> There are two major requirements for the implementation: >>=20 >> 1. all auto-variables that do not have an explicit initializer should = be initialized to >> zero by this option. (Same behavior as CLANG) >>=20 >> 2. keep the current static warning on uninitialized variables = untouched. >>=20 >> In order to satisfy 1, we should check whether an auto-variable has = initializer >> or not; >> In order to satisfy 2, we should add this new transformation after >> "pass_late_warn_uninitialized". >>=20 >> So, we should be able to check whether an auto-variable has = initializer or not after =E2=80=9Cpass_late_warn_uninitialized=E2=80=9D, >> If Not, then insert an initialization for it. >>=20 >> For this purpose, I guess that =E2=80=9CFOR_EACH_LOCAL_DECL=E2=80=9D = might be better? >>=20 >>=20 >> Yes, but do you want to catch variables promoted to register as well >> or just variables >> on the stack? >>=20 >>=20 >> I think both as long as they are source-level auto-variables. Then = which one is better? >>=20 >>=20 >> Another issue is, in order to check whether an auto-variable has = initializer, I plan to add a new bit in =E2=80=9Cdecl_common=E2=80=9D = as: >> /* In a VAR_DECL, this is DECL_IS_INITIALIZED. */ >> unsigned decl_is_initialized :1; >>=20 >> /* IN VAR_DECL, set when the decl is initialized at the declaration. = */ >> #define DECL_IS_INITIALIZED(NODE) \ >> (DECL_COMMON_CHECK (NODE)->decl_common.decl_is_initialized) >>=20 >> set this bit when setting DECL_INITIAL for the variables in FE. then = keep it >> even though DECL_INITIAL might be NULLed. >>=20 >>=20 >> For locals it would be more reliable to set this flag during = gimplification. >>=20 >>=20 >> You mean I can set the flag =E2=80=9CDECL_IS_INITIALIZED (decl)=E2=80=9D= inside the routine =E2=80=9Cgimpley_decl_expr=E2=80=9D (gimplify.c) as = following: >>=20 >> if (VAR_P (decl) && !DECL_EXTERNAL (decl)) >> { >> tree init =3D DECL_INITIAL (decl); >> ... >> if (init && init !=3D error_mark_node) >> { >> if (!TREE_STATIC (decl)) >> { >> DECL_IS_INITIALIZED(decl) =3D 1; >> } >>=20 >> Is this enough for all Frontends? Are there other places that I need = to maintain this bit? >>=20 >>=20 >>=20 >> Do you have any comment and suggestions? >>=20 >>=20 >> As said above - do you want to cover registers as well as locals? >>=20 >>=20 >> All the locals from the source-code point of view should be covered. = (=46rom my study so far, looks like that Clang adds that phase in FE). >> If GCC adds this phase in FE, then the following design requirement >>=20 >> C. The implementation needs to keep the current static warning on = uninitialized >> variables untouched in order to avoid "forking the language=E2=80=9D. >>=20 >> cannot be satisfied. Since gcc=E2=80=99s uninitialized variables = analysis is applied quite late. >>=20 >> So, we have to add this new phase after = =E2=80=9Cpass_late_warn_uninitialized=E2=80=9D. >>=20 >> I'd do >> the actual zeroing during RTL expansion instead since otherwise you >> have to figure youself whether a local is actually used (see = expand_stack_vars) >>=20 >>=20 >> Adding this new transformation during RTL expansion is okay. I will = check on this in more details to see how to add it to RTL expansion = phase. >>=20 >>=20 >> Note that optimization will already made have use of "uninitialized" = state >> of locals so depending on what the actual goal is here "late" may be = too late. >>=20 >>=20 >> This is a really good point=E2=80=A6 >>=20 >> In order to avoid optimization to use the =E2=80=9Cuninitialized=E2=80= =9D state of locals, we should add the zeroing phase as early as = possible (adding it in FE might be best >> for this issue). However, if we have to met the following = requirement: >=20 > So is optimization supposed to pick up zero or is it supposed to act > as if the initializer > is unknown? Good question! Theoretically, the new option -ftrivial-auto-var-init=3Dzero is = supposed to add zero initialization to auto-variables=20 that are not explicitly initialized in order to avoid the possible = undefined behavior.=20 So, I think that with the new option specified, compiler optimization = should pick up zero initialization.=20 Therefore, ideally, zero initializations should be inserted before = optimizations.=20 However, this will conflict with the requirement =E2=80=9C keep the = current static warning on uninitialized variables untouched in order to avoid "forking the language=E2=80=9D." >> C. The implementation needs to keep the current static warning on = uninitialized >> variables untouched in order to avoid "forking the language=E2=80=9D. >>=20 >> We have to move the new phase after all the uninitialized analysis is = done in order to avoid =E2=80=9Cforking the language=E2=80=9D. >>=20 >> So, this is a problem that is not easy to resolve. >=20 > Indeed, those are conflicting goals. Yes, this is the most difficult part for this task.=20 Not sure how CLANG resolved this issue? >=20 >> Do you have suggestion on this? >=20 > No, not any easy ones. Doing more of the uninit analysis early (there > is already an early > uninit pass) which would mean doing IPA analysis turing GCC into more > of a static analysis > tool. Theres the analyzer now, not sure if that can employ an early > LTO phase for example. You mean to enhance =E2=80=9Cpass_early_warn_uninitialized=E2=80=9D or = =E2=80=9Cpass_analyzer=E2=80=9D to catch more uninitialized cases, then add the new =E2=80=9Czero = initialization=E2=80=9D after these passes? However, both =E2=80=9Cpass_early_warn_uninitialized=E2=80=9D and = =E2=80=9Cpass_analyzer=E2=80=9D still utilize some early ipa optimizations. These early optimizations still act as the = initializers are unknown.=20 So, looks like the conflicting cannot be completely resolved.=20 Another thought, If we still add the initializations at = =E2=80=9Cpass_expand=E2=80=9D as you suggested in the previous email,=20 GCC will be split into two parts, the earlier part before = =E2=80=9Cpass_expand=E2=80=9D all act without the zero initialization And report the uninitialized warnings based on this.=20 The later part after =E2=80=9Cpass_expand=E2=80=9D will pick up zero = initializations. All the RTL optimizations will be applied on the program with all new zero initializations.=20 Will such approach have any potential big issue? Qing >=20 > Richard. >=20 >> Qing >>=20 >>=20 >> Richard. >>=20 >>=20 >> Thanks a lot for the help. >>=20 >> Qing >>=20 >> Richard. >>=20 >>=20 >> Thanks. >>=20 >> Qing