public inbox for gcc-bugs@sourceware.org help / color / mirror / Atom feed
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type [not found] <bug-12288-4@http.gcc.gnu.org/bugzilla/> @ 2013-05-20 14:38 ` paolo.carlini at oracle dot com 2013-05-20 17:15 ` paolo.carlini at oracle dot com 2013-05-20 17:17 ` redi at gcc dot gnu.org 2 siblings, 0 replies; 7+ messages in thread From: paolo.carlini at oracle dot com @ 2013-05-20 14:38 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 Paolo Carlini <paolo.carlini at oracle dot com> changed: What |Removed |Added ---------------------------------------------------------------------------- CC|gcc-bugs at gcc dot gnu.org | Assignee|unassigned at gcc dot gnu.org |paolo.carlini at oracle dot com --- Comment #5 from Paolo Carlini <paolo.carlini at oracle dot com> --- Given that Comment#2 is now fine, I think that the only remaining issue here is that for Comment#1 we also give (additionally to the sane error message) 12288_1.C:4:27: error: ISO C++ forbids declaration of ‘parameter’ with no type [-fpermissive] explicit S (const X::T&) {} Should be easy to avoid, because there is already a mechanism for that in grokdeclarator (type_was_error_mark_node) >From gcc-bugs-return-422650-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org Mon May 20 14:40:18 2013 Return-Path: <gcc-bugs-return-422650-listarch-gcc-bugs=gcc.gnu.org@gcc.gnu.org> Delivered-To: listarch-gcc-bugs@gcc.gnu.org Received: (qmail 9125 invoked by alias); 20 May 2013 14:40:18 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: <gcc-bugs.gcc.gnu.org> List-Archive: <http://gcc.gnu.org/ml/gcc-bugs/> List-Post: <mailto:gcc-bugs@gcc.gnu.org> List-Help: <mailto:gcc-bugs-help@gcc.gnu.org> Sender: gcc-bugs-owner@gcc.gnu.org Delivered-To: mailing list gcc-bugs@gcc.gnu.org Received: (qmail 9059 invoked by uid 48); 20 May 2013 14:40:09 -0000 From: "olegendo at gcc dot gnu.org" <gcc-bugzilla@gcc.gnu.org> To: gcc-bugs@gcc.gnu.org Subject: [Bug target/57339] New: [SH] Wrong ISR FPU register save/restore Date: Mon, 20 May 2013 14:40:00 -0000 X-Bugzilla-Reason: CC X-Bugzilla-Type: new X-Bugzilla-Watch-Reason: None X-Bugzilla-Product: gcc X-Bugzilla-Component: target X-Bugzilla-Version: 4.8.0 X-Bugzilla-Keywords: X-Bugzilla-Severity: normal X-Bugzilla-Who: olegendo at gcc dot gnu.org X-Bugzilla-Status: UNCONFIRMED X-Bugzilla-Priority: P3 X-Bugzilla-Assigned-To: unassigned at gcc dot gnu.org X-Bugzilla-Target-Milestone: --- X-Bugzilla-Flags: X-Bugzilla-Changed-Fields: bug_id short_desc product version bug_status bug_severity priority component assigned_to reporter cc cf_gcctarget Message-ID: <bug-57339-4@http.gcc.gnu.org/bugzilla/> Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit X-Bugzilla-URL: http://gcc.gnu.org/bugzilla/ Auto-Submitted: auto-generated MIME-Version: 1.0 X-SW-Source: 2013-05/txt/msg01323.txt.bz2 Content-length: 6051 http://gcc.gnu.org/bugzilla/show_bug.cgi?idW339 Bug ID: 57339 Summary: [SH] Wrong ISR FPU register save/restore Product: gcc Version: 4.8.0 Status: UNCONFIRMED Severity: normal Priority: P3 Component: target Assignee: unassigned at gcc dot gnu.org Reporter: olegendo at gcc dot gnu.org CC: kkojima at gcc dot gnu.org Target: sh*-*-* The following test case extern void foo (void); #pragma interrupt #pragma nosave_low_regs void isr (void) { foo (); } with -O2 -m4 -ml compiles to: isr: fmov.s fr0,@-r15 fmov.s fr1,@-r15 fmov.s fr2,@-r15 fmov.s fr3,@-r15 fmov.s fr4,@-r15 fmov.s fr5,@-r15 fmov.s fr6,@-r15 fmov.s fr7,@-r15 fmov.s fr8,@-r15 fmov.s fr9,@-r15 fmov.s fr10,@-r15 fmov.s fr11,@-r15 sts.l mach,@-r15 sts.l macl,@-r15 sts.l fpul,@-r15 sts.l fpscr,@-r15 sts.l pr,@-r15 mov.l .L3,r1 lds.l @r1+,fpscr ! 57 fpu_switch/2 [length = 2] mov.l .L4,r1 jsr @r1 nop lds.l @r15+,pr lds.l @r15+,fpscr ! 38 fpu_switch/2 [length = 2] lds.l @r15+,fpul lds.l @r15+,macl lds.l @r15+,mach fmov.s @r15+,fr11 fmov.s @r15+,fr10 fmov.s @r15+,fr9 fmov.s @r15+,fr8 fmov.s @r15+,fr7 fmov.s @r15+,fr6 fmov.s @r15+,fr5 fmov.s @r15+,fr4 fmov.s @r15+,fr3 fmov.s @r15+,fr2 fmov.s @r15+,fr1 rte fmov.s @r15+,fr0 The FPU register save/restore insns in the ISR epilogue and prologue are wrong since they refer to the FPSCR setting at the point of the interruption, but the FPSCR setting is not known at ISR function entry. For example, if the interrupted context's FPSCR has FPSCR.SZ = 1 (64 bit fmov) the code above will not work properly and is most likely going to result in an address error. A better way would be to do something like... isr: ! FPSCR.SZ, FPSCR.PR, FPSCR.FR of interrupted ! context is not known. mov r15,r0 or #7,r0 xor #7,r0 mov r0,r15 ! r15 = 64 bit aligned stack ptr mova .Lvariables,r0 sts.l fpscr,@-r15 ! push FPSCR of interrupted context add #-4,r15 ! keep r15 64 bit aligned lds.l @r0+,fpscr ! FPSCR.SZ = 1, FPSCR.PR = 0, FPSCR.FR = 0 ! if FPSCR.FR was 1 we have now switched the ! FP reg banks and will be saving XF regs. ! that's OK if the ISR code does not switch ! FPSCR.FR. fmov.d dr0,@-r15 fmov.d dr2,@-r15 fmov.d dr4,@-r15 fmov.d dr6,@-r15 fmov.d dr8,@-r15 fmov.d dr10,@-r15 sts.l fpscr,@-r15 ! push FPSCR setting for restoring FP regs ! before ISR returns. lds.l @r0+,fpscr ! default FPSCR setting for ISRs ! FPSCR.PR must be kept = 0 sts.l fpul,@-r15 sts.l mach,@-r15 sts.l macl,@-r15 sts.l pr,@-r15 ... < ISR code > ... lds.l @r15+,pr lds.l @r15+,macl lds.l @r15+,mach lds.l @r15+,fpul lds.l @r15+,fpscr ! pop FPSCR setting for FP reg restore ! FPSCR.SZ = 1, FPSCR.PR = 0, FPSCR.FR = 0 fmov.d @r15+,dr10 fmov.d @r15+,dr8 fmov.d @r15+,dr6 fmov.d @r15+,dr4 fmov.d @r15+,dr2 fmov.d @r15+,dr0 add #4,r15 rte lds.l @r15+,fpscr ! restore FPSCR of interrupted context ! this potentially switches FPSCR.FR .align 2 .Lvariables: .long 1 << 20 ! FPSCR.SZ = 1, FPSCR.PR = 0, FPSCR.FR = 0 .long <default FPSCR value for ISR> ! FPSCR.FR must be = 0 Some remarks: --------- This insn: add #-4,r15 ! keep r15 64 bit aligned can be replaced with one of the push insns that follow the fmov.d pushes. --------- The above won't work for SH2E and SH3E. Those need their own sequence. --------- The stack alignment can be omitted if it has been aligned already somehow else, e.g. by specifying 'sp_switch' function attribute. On SH3* and SH4* the ISR vectors are grouped, so there is always some common code that is executed before it invokes the actual ISR. That code might as well align the stack. --------- An alternative stack alignment sequence could be: mova .Lvariables,r0 mov.l @r0+,r1 and r1,r15 ... .align 2 .Lvariables: .long 0xFFFFFFF8 .long 1 << 20 ! FPSCR.SZ = 1, FPSCR.PR = 0, FPSCR.FR = 0 .long <default FPSCR value for ISR> ! FPSCR.FR must be = 0 However, this is probably a bit slower. --------- On SH4, if the ISR code path changes the FPSCR.FR setting, the FP regs will get scrambled. ISR code must not modify the FPSCR.FR setting. This restriction could be removed by also pushing/popping XF regs (additional 8 push + 8 pop insns). --------- On SH2A and SH2E R0 is not a banked register and must be pushed before dealing with the FP regs. I'm not sure how this should be implemented in the compiler, since currently there are some issues with FP mode switching and fmov.d support, which probably would need to be fixed first. One idea for now would be to emit fixed ISR prologue / epilogue asm blocks that deal with the FP regs, if FP regs need to be saved/restored for an ISR. Kaz, do you happen to have any suggestions/advices/opinions? ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type [not found] <bug-12288-4@http.gcc.gnu.org/bugzilla/> 2013-05-20 14:38 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type paolo.carlini at oracle dot com @ 2013-05-20 17:15 ` paolo.carlini at oracle dot com 2013-05-20 17:17 ` redi at gcc dot gnu.org 2 siblings, 0 replies; 7+ messages in thread From: paolo.carlini at oracle dot com @ 2013-05-20 17:15 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 Paolo Carlini <paolo.carlini at oracle dot com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEW |RESOLVED Known to work| |4.8.0, 4.9.0 Resolution|--- |FIXED --- Comment #6 from Paolo Carlini <paolo.carlini at oracle dot com> --- Spurious error message avoided for 4.9.0. ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type [not found] <bug-12288-4@http.gcc.gnu.org/bugzilla/> 2013-05-20 14:38 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type paolo.carlini at oracle dot com 2013-05-20 17:15 ` paolo.carlini at oracle dot com @ 2013-05-20 17:17 ` redi at gcc dot gnu.org 2 siblings, 0 replies; 7+ messages in thread From: redi at gcc dot gnu.org @ 2013-05-20 17:17 UTC (permalink / raw) To: gcc-bugs http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 --- Comment #7 from Jonathan Wakely <redi at gcc dot gnu.org> --- Excellent, thanks! ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] New: "expected function-definition" error compiling constructor @ 2003-09-15 17:41 redi at gcc dot gnu dot org 2003-09-15 19:45 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type bangerth at dealii dot org ` (3 more replies) 0 siblings, 4 replies; 7+ messages in thread From: redi at gcc dot gnu dot org @ 2003-09-15 17:41 UTC (permalink / raw) To: gcc-bugs PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 Summary: "expected function-definition" error compiling constructor Product: gcc Version: 3.4 Status: UNCONFIRMED Severity: normal Priority: P2 Component: c++ AssignedTo: unassigned at gcc dot gnu dot org ReportedBy: redi at gcc dot gnu dot org CC: gcc-bugs at gcc dot gnu dot org GCC host triplet: i386-unknown-freebsd4.8 The code below fails to compile with CVS g++ from 20030826 (I would have used a more recent build but I can't bootstrap right now). This code compiles without error on Comeau. The GCC config info, code and error follow: $ g++3 -v Reading specs from /data/development/jw/gcc3/bin/../lib/gcc/i386-unknown-freebsd4.8/3.4/specs Configured with: /home/jw/src/cvs/gcc/configure --prefix=/home/jw/gcc3/ --enable-languages=c,c++ --enable-concept-checks --enable-libstdcxx-pch : (reconfigured) /home/jw/src/cvs/gcc/configure --prefix=/home/jw/gcc3/ --enable-languages=c,c++ --enable-concept-checks --enable-libstdcxx-pch : (reconfigured) Thread model: posix gcc version 3.4 20030826 (experimental) $ cat gcc_bug.cc #include <map> struct Foo {}; typedef std::multimap<unsigned short,Foo> FooMap; class SameFoo { public: explicit SameFoo(const FooMap::data_type& lhs) : m_arg(lhs) {} private: const Foo& m_arg; }; $ g++3 gcc_bug.cc gcc_bug.cc:10: error: ISO C++ forbids declaration of `SameFoo' with no type gcc_bug.cc:10: error: only declarations of constructors can be `explicit' gcc_bug.cc:10: error: expected function-definition ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type 2003-09-15 17:41 [Bug c++/12288] New: "expected function-definition" error compiling constructor redi at gcc dot gnu dot org @ 2003-09-15 19:45 ` bangerth at dealii dot org 2003-09-16 11:04 ` redi at gcc dot gnu dot org ` (2 subsequent siblings) 3 siblings, 0 replies; 7+ messages in thread From: bangerth at dealii dot org @ 2003-09-15 19:45 UTC (permalink / raw) To: gcc-bugs PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 bangerth at dealii dot org changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |gdr at gcc dot gnu dot org Status|UNCONFIRMED |NEW Ever Confirmed| |1 Keywords| |diagnostic Last reconfirmed|0000-00-00 00:00:00 |2003-09-15 18:18:23 date| | Summary|"expected function- |unintelligible error message |definition" error compiling |for constructor declaration |constructor |with misspelled type ------- Additional Comments From bangerth at dealii dot org 2003-09-15 18:18 ------- If this code compiles with Comeau, then only because their standard library is non-conforming: there is no type data_type in multimap. Thus, the whole thing boils down to a snippet like this: ---------------------- class X {}; struct S { explicit S (const X::T&) {} }; ---------------------- for which we get this error message: g/x> /home/bangerth/bin/gcc-3.4*/bin/c++ -c y.cc y.cc:4: error: ISO C++ forbids declaration of `S' with no type y.cc:4: error: only declarations of constructors can be `explicit' y.cc:4: error: expected function-definition That's totally unintelligible if one doesn't already know the cause of the problem. Previous versions of gcc gave g/x> /home/bangerth/bin/gcc-3.3*/bin/c++ -c y.cc y.cc:4: error: parse error before `&' token y.cc:4: error: missing ';' before right brace which is not better either. W. ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type 2003-09-15 17:41 [Bug c++/12288] New: "expected function-definition" error compiling constructor redi at gcc dot gnu dot org 2003-09-15 19:45 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type bangerth at dealii dot org @ 2003-09-16 11:04 ` redi at gcc dot gnu dot org 2003-12-29 0:31 ` pinskia at gcc dot gnu dot org 2004-10-28 2:08 ` pinskia at gcc dot gnu dot org 3 siblings, 0 replies; 7+ messages in thread From: redi at gcc dot gnu dot org @ 2003-09-16 11:04 UTC (permalink / raw) To: gcc-bugs PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 ------- Additional Comments From redi at gcc dot gnu dot org 2003-09-16 08:43 ------- Ah yes, I was a bit hasty filing this, should have checked a better reference. I blame the unintelligible diagnostic for the fact I didn't spot that ;-) When not in a constructor the same construct causes a different, but almost as unintelligible message: ----------------- class X {}; typedef X::T xt; ------------------ $ g++3 -c 12288-2.cc map2.cc:3: error: expected init-declarator map2.cc:3: error: expected `,' or `;' This is a bit closer to the message in previous versions (which I don't think is as bad as the current one) but still not much help for the average user trying to work out what's wrong. ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type 2003-09-15 17:41 [Bug c++/12288] New: "expected function-definition" error compiling constructor redi at gcc dot gnu dot org 2003-09-15 19:45 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type bangerth at dealii dot org 2003-09-16 11:04 ` redi at gcc dot gnu dot org @ 2003-12-29 0:31 ` pinskia at gcc dot gnu dot org 2004-10-28 2:08 ` pinskia at gcc dot gnu dot org 3 siblings, 0 replies; 7+ messages in thread From: pinskia at gcc dot gnu dot org @ 2003-12-29 0:31 UTC (permalink / raw) To: gcc-bugs ------- Additional Comments From pinskia at gcc dot gnu dot org 2003-12-28 23:11 ------- Better error message now on the mainline (not good enough though) pr12288.2.cc:5: error: expected unqualified-id before '&' token pr12288.2.cc:5: error: expected `,' or `...' before '&' token pr12288.2.cc:5: error: ISO C++ forbids declaration of `parameter' with no type -- What |Removed |Added ---------------------------------------------------------------------------- Last reconfirmed|2003-12-06 08:28:01 |2003-12-28 23:11:16 date| | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 ^ permalink raw reply [flat|nested] 7+ messages in thread
* [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type 2003-09-15 17:41 [Bug c++/12288] New: "expected function-definition" error compiling constructor redi at gcc dot gnu dot org ` (2 preceding siblings ...) 2003-12-29 0:31 ` pinskia at gcc dot gnu dot org @ 2004-10-28 2:08 ` pinskia at gcc dot gnu dot org 3 siblings, 0 replies; 7+ messages in thread From: pinskia at gcc dot gnu dot org @ 2004-10-28 2:08 UTC (permalink / raw) To: gcc-bugs ------- Additional Comments From pinskia at gcc dot gnu dot org 2004-10-28 02:08 ------- The example in comment#2 is fixed on the mainline (the q part is a different problem, I will file it soon): pr12288.cc:3: error: qT1 in class 'X' does not name a type But other than that we still have problem. -- What |Removed |Added ---------------------------------------------------------------------------- Last reconfirmed|2004-07-29 04:59:21 |2004-10-28 02:08:36 date| | http://gcc.gnu.org/bugzilla/show_bug.cgi?id=12288 ^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2013-05-20 17:17 UTC | newest] Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- [not found] <bug-12288-4@http.gcc.gnu.org/bugzilla/> 2013-05-20 14:38 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type paolo.carlini at oracle dot com 2013-05-20 17:15 ` paolo.carlini at oracle dot com 2013-05-20 17:17 ` redi at gcc dot gnu.org 2003-09-15 17:41 [Bug c++/12288] New: "expected function-definition" error compiling constructor redi at gcc dot gnu dot org 2003-09-15 19:45 ` [Bug c++/12288] unintelligible error message for constructor declaration with misspelled type bangerth at dealii dot org 2003-09-16 11:04 ` redi at gcc dot gnu dot org 2003-12-29 0:31 ` pinskia at gcc dot gnu dot org 2004-10-28 2:08 ` pinskia at gcc dot gnu dot org
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).