From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 70298 invoked by alias); 18 Oct 2015 05:42:41 -0000 Mailing-List: contact gcc-help-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-help-owner@gcc.gnu.org Received: (qmail 70281 invoked by uid 89); 18 Oct 2015 05:42:40 -0000 Authentication-Results: sourceware.org; auth=none X-Virus-Found: No X-Spam-SWARE-Status: Yes, score=6.2 required=5.0 tests=AWL,BAYES_99,BAYES_999,FREEMAIL_FROM,KAM_ASCII_DIVIDERS,RCVD_IN_DNSWL_LOW,SPF_PASS,TBC,T_RP_MATCHES_RCVD autolearn=no version=3.3.2 X-HELO: smtp.laposte.net Received: from smtpoutz13.laposte.net (HELO smtp.laposte.net) (194.117.213.172) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with (AES256-SHA encrypted) ESMTPS; Sun, 18 Oct 2015 05:42:36 +0000 Received: from smtp.laposte.net (localhost [127.0.0.1]) by lpn-prd-vrout001 (Postfix) with ESMTP id 249294E401D4 for ; Sun, 18 Oct 2015 07:42:33 +0200 (CEST) Received: from smtp.laposte.net (localhost [127.0.0.1]) by lpn-prd-vrout001 (Postfix) with ESMTP id 074564E47777 for ; Sun, 18 Oct 2015 07:42:33 +0200 (CEST) Received: from lpn-prd-mstr067.laposte (lpn-prd-mstr067.laposte [10.128.59.68]) by lpn-prd-vrout001 (Postfix) with ESMTP id E20A94E401D4 for ; Sun, 18 Oct 2015 07:42:32 +0200 (CEST) Date: Sun, 18 Oct 2015 05:42:00 -0000 From: john.frankish@laposte.net To: gcc-help Message-ID: <1551573086.245176.1445146952715.JavaMail.zimbra@laposte.net> In-Reply-To: References: <54BA30F3.4080202@sunrus.com.cn> <55E27DB9.6050606@hotmail.com> <560CB2C0.7000107@hotmail.com> Subject: Re: [Consult] tile: How to construct testsuite for tile architecture without hardware. MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable X-VR-SrcIP: [91.74.228.192] X-VR-FullState: 0 X-VR-Score: 0 X-VR-Cause-1: gggruggvucftvghtrhhoucdtuddrfeekgedrleefgddvfeefucetufdoteggodetrfcurfhrohhfihhl X-VR-Cause-2: vgemucfntefrqffuvffgnecuuegrihhlohhuthemucehtddtnecunecujfgurhepfffhvffkjghfufgg X-VR-Cause-3: tgfgihfothesthhqtgdtredtjeenucfhrhhomhepjhhohhhnrdhfrhgrnhhkihhshheslhgrphhoshht X-VR-Cause-4: vgdrnhgvthenucffohhmrghinhepthhilhgvrhgrrdgtohhmnecurfgrrhgrmhepmhhouggvpehsmhht X-VR-Cause-5: phhouhhtpdhhvghloheplhhpnhdqphhrugdqmhhsthhrtdeijedrphhrohhsohguihgvpdhinhgvthep X-VR-Cause-6: uddtrdduvdekrdehledrieekpdhmrghilhhfrhhomhepjhhohhhnrdhfrhgrnhhkihhshheslhgrphho X-VR-Cause-7: shhtvgdrnhgvthdprhgtphhtthhopehgtggtqdhhvghlphesghgttgdrghhnuhdrohhrgh X-VR-AvState: No X-VR-State: 0 X-IsSubscribed: yes X-SW-Source: 2015-10/txt/msg00115.txt.bz2 ----- Mail original -----=20 De: "Chen Gang" =20 =C3=80: "Chris Metcalf" , walt@tilera.com=20 Cc: "Jeff Law" , "gcc-help" , "Mike S= tump" , "Michael Eager" , "Richard= Henderson" , "Peter Maydell" , "= Andreas F=C3=A4rber" , rth@twiddle.net=20 Envoy=C3=A9: Dimanche 18 Octobre 2015 09:23:45=20 Objet: Re: [Consult] tile: How to construct testsuite for tile architecture= without hardware.=20 Hello all:=20 After fix one qemu issue about floating point insns, correct the related=20 environments (make some symbol links), and reconfig building options (so=20 support g++), the current gcc testsuite result are:=20 =3D=3D=3D gcc Summary =3D=3D=3D=20 # of expected passes 78156=20 # of unexpected failures 98=20 # of unexpected successes 1=20 # of expected failures 120=20 # of unresolved testcases 1=20 # of unsupported tests 1384=20 =3D=3D=3D g++ Summary =3D=3D=3D=20 # of expected passes 83999=20 # of unexpected failures 78=20 # of unexpected successes 2=20 # of expected failures 277=20 # of unsupported tests 3568=20 [root@localhost build-gcc-tile]# gcc/xgcc -v=20 Using built-in specs.=20 COLLECT_GCC=3Dgcc/xgcc=20 Target: tilegx-gchen-linux=20 Configured with: ../gcc/configure --target=3Dtilegx-gchen-linux --disable-n= ls --enable-languages=3Dc,c++ --enable-threads --enable-shared --enable-lib= ssp --prefix=3D/upstream/release-tile --with-headers=3D/upstream/release-ti= le/include --with-sysroot=3D/upstream/release-tile --disable-multilib --dis= able-libquadmath --disable-libgomp --disable-libatomic=20 Thread model: posix=20 gcc version 6.0.0 20151012 (experimental) (GCC)=20 [root@localhost build-gcc-tile]# gcc/xg++ -v=20 Using built-in specs.=20 COLLECT_GCC=3Dgcc/xg++=20 Target: tilegx-gchen-linux=20 Configured with: ../gcc/configure --target=3Dtilegx-gchen-linux --disable-n= ls --enable-languages=3Dc,c++ --enable-threads --enable-shared --enable-lib= ssp --prefix=3D/upstream/release-tile --with-headers=3D/upstream/release-ti= le/include --with-sysroot=3D/upstream/release-tile --disable-multilib --dis= able-libquadmath --disable-libgomp --disable-libatomic=20 Thread model: posix=20 gcc version 6.0.0 20151012 (experimental) (GCC)=20 Welcome any ideas, suggestions, and completions.=20 Thanks.=20 On 10/17/15 14:39, Chen Gang wrote:=20 >=20 > After rebuild gcc/binutils with the latest code and correct the building= =20 > config parameters, the gcc testsuite result is:=20 >=20 > =3D=3D=3D gcc Summary =3D=3D=3D=20 >=20 > # of expected passes 77855=20 > # of unexpected failures 162=20 > # of unexpected successes 1=20 > # of expected failures 120=20 > # of unresolved testcases 131=20 > # of unsupported tests 1477=20 >=20 > root@localhost tile]# /upstream/release-tile/bin/tilegx-gchen-linux-gcc -= v=20 > Using built-in specs.=20 > COLLECT_GCC=3D/upstream/release-tile/bin/tilegx-gchen-linux-gcc=20 > COLLECT_LTO_WRAPPER=3D/upstream/release-tile/libexec/gcc/tilegx-gchen-lin= ux/6.0.0/lto-wrapper=20 > Target: tilegx-gchen-linux=20 > Configured with: ../gcc/configure --target=3Dtilegx-gchen-linux --disable= -nls --enable-languages=3Dc --enable-threads --enable-shared --enable-libss= p --prefix=3D/upstream/release-tile --with-headers=3D/upstream/release-tile= /include --with-sysroot=3D/upstream/release-tile --disable-multilib=20 > Thread model: posix=20 > gcc version 6.0.0 20151012 (experimental) (GCC)=20 >=20 > Welcome any members' ideas, suggestions, and completions.=20 >=20 >=20 > Thanks.=20 >=20 >=20 > On 10/5/15 19:59, Chen Gang wrote:=20 >> After fix the tillegx qemu's fsingle floating point insns issue, the gcc= test suite result is:=20 >>=20 >> =3D=3D=3D gcc Summary =3D=3D=3D=20 >>=20 >> # of expected passes 77272=20 >> # of unexpected failures 365=20 >> # of unexpected successes 7=20 >> # of expected failures 113=20 >> # of unresolved testcases 141=20 >> # of unsupported tests 1476=20 >>=20 >> And I shall continue to analyze the left issues, next. :-)=20 >>=20 >>=20 >> Thanks.=20 >> --=20 >> Chen Gang=20 >>=20 >> Open, share, and attitude like air, water, and life which God blessed=20 >>=20 >>=20 >> ----------------------------------------=20 >>> From: xili_gchen_5257@hotmail.com=20 >>> To: cmetcalf@ezchip.com; walt@tilera.com=20 >>> CC: law@redhat.com; gcc-help@gcc.gnu.org; mikestump@comcast.net; eager@= eagerm.com; rth@redhat.com; peter.maydell@linaro.org; afaerber@suse.de; rth= @twiddle.net=20 >>> Subject: RE: [Consult] tile: How to construct testsuite for tile archit= ecture without hardware.=20 >>> Date: Sun, 4 Oct 2015 19:55:44 +0800=20 >>>=20 >>> After finish additional 7 qemu patches, our tilegx qemu can test the gc= c=20 >>> testsuite successfully (no any unimplementation issues). The test resul= t=20 >>> is:=20 >>>=20 >>> =3D=3D=3D gcc Summary =3D=3D=3D=20 >>>=20 >>> # of expected passes 77012=20 >>> # of unexpected failures 622=20 >>> # of unexpected successes 7=20 >>> # of expected failures 113=20 >>> # of unresolved testcases 143=20 >>> # of unsupported tests 1476=20 >>>=20 >>>=20 >>> For the 622 left issues:=20 >>>=20 >>> - Maybe still tilegx implementation issue, e.g.=20 >>>=20 >>> the floating point single mul instruction cann't get the result which=20 >>> tilegx gcc expected (it is only related with calculating: float32_mul=20 >>> calculation result is not match tilegx gcc expected).=20 >>>=20 >>> - Maybe envorintments configuration issues, e.g.=20 >>>=20 >>> for gcc guality_check, it may call gdb outside, so I need config the=20 >>> related gdb correctlly.=20 >>>=20 >>> - Maybe tilegx gcc issues, e.g.=20 >>>=20 >>> for some UNKNOWN instructions, probably generated by incorrect jump=20 >>> instructions, one sample is "-fpic -mcmodel=3Dlarge" for gcc nested-5=20 >>> test.=20 >>>=20 >>> Next, I shall:=20 >>>=20 >>> - continue to fix the left 622 issues, and send related patches to=20 >>> qemu-devel mailing list or gcc-patches mailing list.=20 >>>=20 >>> - then rewrite floating point instructions, and try to let the related= =20 >>> code merged into qemu mainline with qemu members' help (at present,=20 >>> floating point implementation is a temporary implementation).=20 >>>=20 >>> - at last, try to finish all left unimplemented instructions (there are= =20 >>> still some unimplement insns left, but at present, gcc testsuite does=20 >>> not use them), and start tilegx qemu system mode, next.=20 >>>=20 >>>=20 >>> Welcome any members' ideas, suggestion, and completions. :-)=20 >>>=20 >>>=20 >>> Thanks.=20 >>>=20 >>> Chen Gang=20 >>>=20 >>> Open, share, and attitude like air, water, and life which God blessed=20 >>>=20 >>>=20 >>> ----------------------------------------=20 >>>> Date: Thu, 1 Oct 2015 12:12:48 +0800=20 >>>> From: xili_gchen_5257@hotmail.com=20 >>>> To: cmetcalf@ezchip.com; walt@tilera.com=20 >>>> CC: law@redhat.com; gcc-help@gcc.gnu.org; mikestump@comcast.net; eager= @eagerm.com; rth@redhat.com; peter.maydell@linaro.org; afaerber@suse.de; rt= h@twiddle.net=20 >>>> Subject: Re: [Consult] tile: How to construct testsuite for tile archi= tecture without hardware.=20 >>>>=20 >>>>=20 >>>> At present, tilegx is integrated into qemu master branch successfully= =20 >>>> with the much help by qemu members.=20 >>>>=20 >>>> But sorry, I did not finish gcc testsuite in time, the main excuse is= =20 >>>>=20 >>>> - I am busy on the normal work -- sometimes, have to do overtime work.= =20 >>>> (my company started up an urgent project, but couldn't find suitable=20 >>>> persons for it in short time, so I have to do it during these days).=20 >>>>=20 >>>> Next, in Chinese National Day, I have enough free time resources on op= en=20 >>>> source, hope I finish our gcc testsuite within Chinese National Day (i= t=20 >>>> is from 2015-10-01 -- 2015-10-07).=20 >>>>=20 >>>> Thanks.=20 >>>>=20 >>>> On 8/30/15 11:50, Chen Gang wrote:=20 >>>>>=20 >>>>> Oh, sorry again, I did not finish qemu test within this month, again.= =20 >>>>> Current status is:=20 >>>>>=20 >>>>> =3D=3D=3D gcc Summary =3D=3D=3D=20 >>>>>=20 >>>>> # of expected passes 76725=20 >>>>> # of unexpected failures 790=20 >>>>> # of unexpected successes 7=20 >>>>> # of expected failures 113=20 >>>>> # of unresolved testcases 142=20 >>>>> # of unsupported tests 1483=20 >>>>>=20 >>>>>=20 >>>>> My main excuses are:=20 >>>>>=20 >>>>> - We are lack of tilegx floating point insns related documents, we ha= ve=20 >>>>> to 'guess' implementation details. At last, we found ways to process= =20 >>>>> it (we spent more than a half month free time resources for it).=20 >>>>>=20 >>>>> - My related qemu code need be refactor before merging into qemu mast= er=20 >>>>> tree. At present, we are refactoring and merging tilegx code to qemu= =20 >>>>> master tree (mainly by qemu related members).=20 >>>>>=20 >>>>> Hope I can finish qemu tilegx gcc testsuite within next month (I gues= s,=20 >>>>> it should be done in time).=20 >>>>>=20 >>>>> - Test the refactored and merged code again, based on it, I continue = to=20 >>>>> fix gcc testsuite related issues.=20 >>>>>=20 >>>>> - Fix all 700+ gcc testsuite issues (10+ for pending issues, and the= =20 >>>>> others are for UN-implementation insns issues, I guess).=20 >>>>>=20 >>>>>=20 >>>>> Thanks.=20 >>>>>=20 >>>>>=20 >>>>> On 7/30/15 05:01, Chen Gang wrote:=20 >>>>>> On 7/20/15 06:26, Chen Gang wrote:=20 >>>>>>> Hello all:=20 >>>>>>>=20 >>>>>>> At present (after a half year), with the much help of QEMU members,= =20 >>>>>>> tilegx linux-user can work (manually using busybox's sh, ls, cp, mv= , vi=20 >>>>>>> are OK).=20 >>>>>>>=20 >>>>>>> Next, I shall start gcc testsuite with qemu tilegx linux-user, hope= I=20 >>>>>>> can finish within this month (2015-07-31).=20 >>>>>>>=20 >>>>>>=20 >>>>>> Sorry, I can not finish the qemu test within this month. The reason = is=20 >>>>>> my tilegx qemu is still lack of some instructions implementation, ne= xt,=20 >>>>>> I should finish them firstly (I implemented most insns, but not all)= .=20 >>>>>>=20 >>>>>> =3D=3D=3D gcc Summary =3D=3D=3D=20 >>>>>>=20 >>>>>> # of expected passes 69830=20 >>>>>> # of unexpected failures 4312=20 >>>>>> # of unexpected successes 2=20 >>>>>> # of expected failures 108=20 >>>>>> # of unresolved testcases 1776=20 >>>>>> # of unsupported tests 1567=20 >>>>>>=20 >>>>>> More than 4K failures are because tilegx qemu lacks of instruction's= =20 >>>>>> implementation, and also more than 10 issues are for TIMEOUT (I gues= s,=20 >>>>>> my original tilegx qemu instruction's implementation still has bugs)= .=20 >>>>>>=20 >>>>>> I shall try to finish it within next month, although I am not quite = sure=20 >>>>>> (it needs enough time resources on it, but I am lack of time resourc= es:=20 >>>>>> I can only spend less 1 hour per work day and 2 hours per week end d= ay).=20 >>>>>>=20 >>>>>>=20 >>>>>>=20 >>>>>> Thanks.=20 >>>>>>=20 >>>>>>> Thanks.=20 >>>>>>>=20 >>>>>>> On 1/24/15 21:31, Chen Gang S wrote:=20 >>>>>>>> Hello Maintainers:=20 >>>>>>>>=20 >>>>>>>> The result of gcc testsuite for tile without qemu is below, I gues= s the=20 >>>>>>>> environments construction is OK:=20 >>>>>>>>=20 >>>>>>>> =3D=3D=3D gcc Summary =3D=3D=3D=20 >>>>>>>>=20 >>>>>>>> # of expected passes 66650=20 >>>>>>>> # of unexpected failures 42=20 >>>>>>>> # of unexpected successes 1=20 >>>>>>>> # of expected failures 107=20 >>>>>>>> # of unresolved testcases 16558=20 >>>>>>>> # of unsupported tests 1771=20 >>>>>>>>=20 >>>>>>>> Based on normal testsuite and microblaze testsuite configuration, = it is=20 >>>>>>>> easy to config gcc testsuite for tile. Thank gcc testsuit member a= nd=20 >>>>>>>> microblaze member very much. :-)=20 >>>>>>>>=20 >>>>>>>> - Copy the original microblaze exp file as tile exp file. Export t= ile=20 >>>>>>>> ip to /etc/hosts.=20 >>>>>>>>=20 >>>>>>>> - Change the related compiler name and related patch to tile in ti= le=20 >>>>>>>> exp file.=20 >>>>>>>>=20 >>>>>>>> - Use tile cflag "-m64 -mcpu=3Dtilegx -mcmodel=3Dlarge -mlittle-en= dian"=20 >>>>>>>> instead of the original microblaze cflag, in tile exp file.=20 >>>>>>>>=20 >>>>>>>> If OK, I shall try to make a tile patch within this month. Next mo= nth,=20 >>>>>>>> I shall start tile qemu (I am preparing for it within this month).= After=20 >>>>>>>> communicate with qemu members, I guess, it is possible to finish i= t.=20 >>>>>>>>=20 >>>>>>>>=20 >>>>>>>> Welcome any ideas, suggestions, and completions.=20 >>>>>>>>=20 >>>>>>>> Thanks.=20 >>>>>>>>=20 >>>>>>>> On 1/20/15 10:02, Chen Gang S wrote:=20 >>>>>>>>> On 1/20/15 00:10, Chris Metcalf wrote:=20 >>>>>>>>>> On 1/17/2015 4:52 AM, Chen Gang S wrote:=20 >>>>>>>>>>> Hello Maintainers:=20 >>>>>>>>>>>=20 >>>>>>>>>>> I am trying to make gcc patches for tile, and need testsuite wi= thout=20 >>>>>>>>>>> related hardware, so I consult:=20 >>>>>>>>>>=20 >>>>>>>>>> Unfortunately, at the current time, the simulator is only availa= ble as part of the commercial Multicore Development Environment package. A = qemu port for tile hardware would certainly be welcome, but to my knowledge= , no one has completed such a port yet. ISA and ABI documentation for tileg= x are available at www.tilera.com/scm.=20 >>>>>>>>>>=20 >>>>>>>>>=20 >>>>>>>>> OK, thank you very much, and the documentation is really valuable= to me.=20 >>>>>>>>>=20 >>>>>>>>> After finish gcc tile testsuite (hope I can finish within this mo= nth),=20 >>>>>>>>> I shall continue for tile qemu, I should try my best for it, hope= I can=20 >>>>>>>>> finish: "can run gcc testsuite with the tile qemu".=20 >>>>>>>>>=20 >>>>>>>>> At present, it is really very hard to me, but for me, it is my du= ty (I=20 >>>>>>>>> met it, and it was within my original boarder).=20 >>>>>>>>>=20 >>>>>>>>>=20 >>>>>>>>> For me, nothing is impossible, if God blessed.=20 >>>>>>>>>=20 >>>>>>>>>=20 >>>>>>>>> Thanks.=20 >>>>>>>>>=20 >>>>>>>>=20 >>>>>>>=20 >>>>>>> Thanks.=20 >>>>>>>=20 >>>>>>=20 >>>>>=20 >>>>> --=20 >>>>> Chen Gang=20 >>>>>=20 >>>>> Open, share, and attitude like air, water, and life which God blessed= =20 >>>>>=20 >>>>>=20 >>>>=20 >>>> --=20 >>>> Chen Gang (=E9=99=88=E5=88=9A)=20 >>>>=20 >>>> Open, share, and attitude like air, water, and life which God blessed= =20 >>>=20 >>=20 >>=20 >=20 --=20 Chen Gang (=E9=99=88=E5=88=9A)=20 Open, share, and attitude like air, water, and life which God blessed=20 [test - apologies, but I am unable to post, so maybe I can reply]