From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx0a-001b2d01.pphosted.com (mx0a-001b2d01.pphosted.com [148.163.156.1]) by sourceware.org (Postfix) with ESMTPS id CA0AE3858C2D for ; Tue, 24 Oct 2023 16:09:10 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org CA0AE3858C2D Authentication-Results: sourceware.org; dmarc=none (p=none dis=none) header.from=linux.ibm.com Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=linux.ibm.com ARC-Filter: OpenARC Filter v1.0.0 sourceware.org CA0AE3858C2D Authentication-Results: server2.sourceware.org; arc=none smtp.remote-ip=148.163.156.1 ARC-Seal: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1698163753; cv=none; b=fAUTkeYVJrCI0FcK3aJ/PDCGoHjpJtN+xjkshv36FrV99ncjGxJzQQgOErovN/yOyz8FwjO1MmzZidRoAg1AQC1PhYYtE7XQioRvQ7k9X9Bg3lnMoPBTd+CQhr2Rt8Sotb7QLHf9TipKcsJGBIZBy4N85ZT0Adlay+LmZezFreg= ARC-Message-Signature: i=1; a=rsa-sha256; d=sourceware.org; s=key; t=1698163753; c=relaxed/simple; bh=GoSd+++ClQ+GDQ7P1pPru9nFmc39avW4qeIAAMbW3Ss=; h=DKIM-Signature:Message-ID:Subject:From:To:Date:Mime-Version; b=Jd3Eifngx9rWg0V/xFBsXaRogtSFN3pdMvzLM/sgA0voGiflW0eWAUvtl9VGE0167j/V71T2DV7NTgWlPROiuLT+ulIFb4atMu9eLzFbikOnvbrHwiiX2BOjLPLyCwaPheAPksYvSZUpWRMH7BMPtRmi+P95nLgcWaTIHwtZtmU= ARC-Authentication-Results: i=1; server2.sourceware.org Received: from pps.filterd (m0360083.ppops.net [127.0.0.1]) by mx0a-001b2d01.pphosted.com (8.17.1.19/8.17.1.19) with ESMTP id 39OG7swr021395 for ; Tue, 24 Oct 2023 16:09:09 GMT DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=ibm.com; h=message-id : subject : from : to : date : in-reply-to : references : content-type : mime-version : content-transfer-encoding; s=pp1; bh=oxpNIZjKobUXFAHhkH9UlewpDRFhFcazuFsR1ngipcQ=; b=l/ygzGQWgG38sNPHNuKuA+4znNjvwTWhSPwPhib7Of/9s2f2NAAudsHWopy+dZmjm02X f6jQnSE3lHLcRUvkLjCyro3FnOSkz1KZGYNYIKpG/5eyimiG+Nr3YbtkfkjvJvQnIwdQ FlQiyj9bAv6NlkV9m0/Q/MCZYOUPnqASV+nfoImKX7Ejp0Ae4yYh0+ZRFLjEL7RrUiTe TPEPY1EV+HayxJWKxkBWvv0Ei65jje0H+g30nWkX2oAiEmMw0Nt1vvmWgu2QSnU4n4tn XXnWCrzd8FGiVtyoZ+1Kw0WDchAgwsSozi91gAdE9fn0rJxmquCvryScp3nySjnzUBBa Gw== Received: from pps.reinject (localhost [127.0.0.1]) by mx0a-001b2d01.pphosted.com (PPS) with ESMTPS id 3txh5vg3nb-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT) for ; Tue, 24 Oct 2023 16:09:08 +0000 Received: from m0360083.ppops.net (m0360083.ppops.net [127.0.0.1]) by pps.reinject (8.17.1.5/8.17.1.5) with ESMTP id 39OG8Ebv024179 for ; Tue, 24 Oct 2023 16:08:58 GMT Received: from ppma13.dal12v.mail.ibm.com (dd.9e.1632.ip4.static.sl-reverse.com [50.22.158.221]) by mx0a-001b2d01.pphosted.com (PPS) with ESMTPS id 3txh5vg355-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 24 Oct 2023 16:08:56 +0000 Received: from pps.filterd (ppma13.dal12v.mail.ibm.com [127.0.0.1]) by ppma13.dal12v.mail.ibm.com (8.17.1.19/8.17.1.19) with ESMTP id 39OEqJdH024365; Tue, 24 Oct 2023 16:05:20 GMT Received: from smtprelay07.dal12v.mail.ibm.com ([172.16.1.9]) by ppma13.dal12v.mail.ibm.com (PPS) with ESMTPS id 3tvu6k074t-1 (version=TLSv1.2 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT); Tue, 24 Oct 2023 16:05:20 +0000 Received: from smtpav06.dal12v.mail.ibm.com (smtpav06.dal12v.mail.ibm.com [10.241.53.105]) by smtprelay07.dal12v.mail.ibm.com (8.14.9/8.14.9/NCO v10.0) with ESMTP id 39OG5Ijj50594394 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-GCM-SHA384 bits=256 verify=OK); Tue, 24 Oct 2023 16:05:18 GMT Received: from smtpav06.dal12v.mail.ibm.com (unknown [127.0.0.1]) by IMSVA (Postfix) with ESMTP id 8BA5158059; Tue, 24 Oct 2023 16:05:18 +0000 (GMT) Received: from smtpav06.dal12v.mail.ibm.com (unknown [127.0.0.1]) by IMSVA (Postfix) with ESMTP id D3E6C58055; Tue, 24 Oct 2023 16:05:17 +0000 (GMT) Received: from wecm-9-67-110-146.wecm.ibm.com (unknown [9.67.110.146]) by smtpav06.dal12v.mail.ibm.com (Postfix) with ESMTP; Tue, 24 Oct 2023 16:05:17 +0000 (GMT) Message-ID: <959a0bed35329b8327e16eb3dddd31c27361e29a.camel@linux.ibm.com> Subject: Re: [Patch 1/2] PowerPC, Fix-test-gdb.base-store.exp From: Carl Love To: Andrew Burgess , Carl Love , gdb-patches@sourceware.org, UlrichWeigand Date: Tue, 24 Oct 2023 09:05:17 -0700 In-Reply-To: <87y1fs2yld.fsf@redhat.com> References: <6f9c8fa20962129048384d74f6f15d1b37d255ed.camel@us.ibm.com> <76b8ed7b93608d40ab42b0538319f78eaf7d621c.camel@us.ibm.com> <87bkcyhc5g.fsf@redhat.com> <240a25512793955b83d82d90cd6d4f0ce09c564e.camel@us.ibm.com> <87y1fs2yld.fsf@redhat.com> Content-Type: text/plain; charset="UTF-8" X-Mailer: Evolution 3.28.5 (3.28.5-22.el8) Mime-Version: 1.0 Content-Transfer-Encoding: 7bit X-TM-AS-GCONF: 00 X-Proofpoint-ORIG-GUID: 1_Ow-H7GKmb0nDw-hjv0WsjO9GYeH6bB X-Proofpoint-GUID: uAUwsxQucV7j9tYydg0H40Uy_1n563It X-Proofpoint-Virus-Version: vendor=baseguard engine=ICAP:2.0.272,Aquarius:18.0.980,Hydra:6.0.619,FMLib:17.11.176.26 definitions=2023-10-24_16,2023-10-24_01,2023-05-22_02 X-Proofpoint-Spam-Details: rule=outbound_notspam policy=outbound score=0 mlxlogscore=999 mlxscore=0 clxscore=1015 malwarescore=0 priorityscore=1501 suspectscore=0 spamscore=0 adultscore=0 phishscore=0 bulkscore=0 impostorscore=0 lowpriorityscore=0 classifier=spam adjust=0 reason=mlx scancount=1 engine=8.12.0-2310170001 definitions=main-2310240139 X-Spam-Status: No, score=-11.6 required=5.0 tests=BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_EF,GIT_PATCH_0,RCVD_IN_MSPIKE_H4,RCVD_IN_MSPIKE_WL,SPF_HELO_NONE,SPF_PASS,TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: Andrew: On Tue, 2023-10-24 at 09:50 +0100, Andrew Burgess wrote: > Carl Love writes: > > > Andrew: > > > > Thanks for the review and comments. See responses below. > > Carl, > > Sorry for my slow reply. > > > On Mon, 2023-10-16 at 15:31 +0100, Andrew Burgess wrote: > > > Carl Love writes: > > > > > > > GDB maintainers: > > > > > > > > This is the first patch in the series which fixes the DWWARF > > > > register > > > > mapping and signal handling issues on PowerPC. > > > > > > > > Carl > > > > > > > > ----------------------------------------------- > > > > > > > > rs6000, Fix Linux DWARF register mapping > > > > > > > > The PowerPC DWARF register mapping is the same for the > > > > .eh_frame > > > > and > > > > .debug_frame on Linux. PowerPC uses different mapping for > > > > .eh_frame and > > > > .debug_frame on other operating systems. The current GDB > > > > support > > > > for > > > > mapping the DWARF registers in > > > > rs6000_linux_dwarf2_reg_to_regnum > > > > and > > > > rs6000_adjust_frame_regnum file gdb/rs6000-tdep.c is not > > > > correct > > > > for Linux. > > > > The files have some legacy mappings for spe_acc, spefscr, EV > > > > which > > > > was > > > > removed from GCC in 2017. > > > > > > > > This patch adds a two new functions > > > > rs6000_linux_dwarf2_reg_to_regnum, > > > > and rs6000_linux_adjust_frame_regnum in file gdb/ppc-linux- > > > > tdep.c > > > > to handle > > > > the DWARF register mappings on Linux. Function > > > > rs6000_linux_dwarf2_reg_to_regnum is installed for both > > > > gdb_dwarf_to_regnum > > > > and gdbarch_stab_reg_to_regnum since the mappings are the same. > > > > > > > > The ppc_linux_init_abi function in gdb/ppc-linux-tdep.c is > > > > updated > > > > to > > > > call set_gdbarch_dwarf2_reg_to_regnum map the new function > > > > rs6000_linux_dwarf2_reg_to_regnum for the > > > > architecture. Similarly, > > > > dwarf2_frame_set_adjust_regnum is called to map > > > > rs6000_linux_adjust_frame_regnum into the architecture. > > > > > > > > The second issue fixed by this patch is the check for > > > > subroutine > > > > process_event_stop_test. Need to make sure the frame is not > > > > the > > > > SIGTRAMP_FRAME. The sequence of events on most platforms is: > > > > > > Usually for GDB we avoid bundling unrelated changes into a single > > > commit. Each commit should address one self contained issue (as > > > far > > > as > > > possible). > > > > > > I really struggling to see any connection between the two fixes > > > you > > > have > > > here. > > > > As mentioned in the commit log, the first patch which fixes the > > DWARF > > register mapping fixes two of the 5 failures seen in store.exp. > > Specifically the patch fixes the issue of GDB not stopping in the > > handler which results in the step.exp test failing. The second > > patch > > then fixes the rest of the issues with the step.exp test. > > I'm now even more confused. > > The commit message for the first patch suggests that it contains two > separate fixes, but the commit message, even your updated commit > message > for 'ver2' only mentions store.exp. > > To be more specific, in patch 1/2, the changes in ppc-linux-tdep.c > seem > unrelated to the changes in infrun.c. And I'm confused about how the > changes in infrun.c can fix anything in the store.exp test script -- > the > comments relating to that part of the patch specifically talk about > 'nexti' and signal handlers, but I don't see either of these being > used > in store.exp (we do use 'next' though, so maybe that's what you > mean?) > But I'm not clear if a signal is invoked in that test at all. > > In the text above you do mention step.exp, so maybe that's what the > infrun.c changes are fixing? I'll grab a PPC machine and have a > play, > but your 'ver 2' commit message has no mention of step.exp, so if > that > test is important you might need to update the text. > The 128-bit floating point issues in store.exp will need a Power10 system which supports the 128-bit floating point types. A Power 9 system is not sufficient. OK, I think I see what is missing in the explanation. So initially, I fixed the DWARF register mapping for the store.exp test. This fixed a couple of the issues with the store.exp test. Then in the second patch we fixed the DWARF register mapping for the 128-bit floating point values. Specifically, DWARF used the same register numbers for 64-bit and 128-bit floating point values. However, the 128-bit floating point values are stored in the VSR register file while the 64-bit values are stored in the floating point registers. The function ieee_128_float_regnum_adjust takes care of that fix. So at this point, the store.exp test works. But now I have a regression failure on gdb.base/sigstep.exp. This test is specifically dealing with signal handlers. Fixing the register mapping, in patch 1, exposed an underlying bug in the signal handler code. The fixes for the nexti, stepi in the signal handling fixes the regression error. The signal handling fix is needed as part of the DWARF register mapping fix so we don't have any regression failures and so these fixes got rolled into the first patch but it was not really explained in the original patch description. So yes, there are really two things getting fixed in the first patch. But you don't see the signal handler issue until after the register mapping is fixed for Linux. The signal handler issues should be exposed on Power 9 if we split out the DWARF register mapping from the signal handler fixes. I didn't play with these individually on Power 9 as part of the patch work. All of the patch work was done on Power 10. So maybe it will be best if I redo the two patch sequence into three patches. Patch 1, fix the mapping; Patch 2 fix the signal handling issue exposed once the register mapping is fixed; Patch 3 fix the handling of the 128-bit float DWARF register mapping. This should then make it clearer what is going on in each patch. Thoughts? Carl > Usually in GDB when we talk about each patch containing a single fix > we're talking about changes to GDB functionality, not fixing a test > script. So it's not: 'Patch 1/1 fixes all of store.exp' but rather, > 'Patch 1/2 fixes PPC DWARF register mapping' and 'Patch 2/2 fixes > nexti > blah blah signal frames'. > > > > > 1) Some code is running when a signal arrives. > > > > 2) The kernel handles the signal and dispatches to the > > > > handler. > > > > ... > > > > > > > > However on PowerPC the sequence of events is: > > > > > > > > 1) Some code is running when a signal arrives. > > > > 2) The kernel handles the signal and dispatches to the > > > > trampoline. > > > > 3) The trampoline performs a normal function call to the > > > > handler. > > > > ... > > > > > > > > We want "nexti" to step into, not over, signal handlers invoked > > > > by the kernel. This is the case most platforms as the kernel > > > > puts > > > > a > > > > signal trampoline frame onto the stack to handle proper return > > > > after the > > > > handler. However, on some platforms such as PowerPC, the > > > > kernel > > > > actually > > > > uses a trampoline to handle *invocation* of the handler. > > > > > > > > The issue is fixed in function process_event_stop_test by > > > > adding a > > > > check > > > > that the frame is not a SIGTRAMP_FRAME to the if statement to > > > > stop > > > > in > > > > a subroutine call. This prevents GDB from erroneously > > > > detecting > > > > the > > > > trampoline invocation as a subroutine call. > > > > > > > > This patch fixes two regression test failures in > > > > gdb.base/store.exp. It > > > > also fixes two regression failures in gdb.python/py-thread- > > > > exited.exp. > > > > > > On the one random PPC box I tried this patch on, I'm not seeing > > > any > > > failures in gdb.python/py-thread-exited.exp either before, or > > > after > > > this > > > commit. > > > > > > Which tests in gdb.python/py-thread-exited.exp are you seeing as > > > broken? > > > And which of the two fixes in this commit fix the problems you're > > > seeing? > > > > The comment about the py-thread-exited.exp should have been > > removed. I > > found that sometimes that test fails and sometime passes. I have > > yet > > to dig into it and figure out why the test is inconsistent. The > > inconsistent behavior exists with and without these patches. > > > > I will remove the comment about the gdb.python/py-thread-exited.exp > > fixes as that is erroneous. > > Thanks. > > > > > > > Patch has been tested on Power 8 LE/BE, Power 9 LE/BE, Power 10 > > > > with no > > > > new regressions. > > > > --- > > > > gdb/infrun.c | 13 ++++++++++ > > > > gdb/ppc-linux-tdep.c | 56 > > > > ++++++++++++++++++++++++++++++++++++++++++++ > > > > 2 files changed, 69 insertions(+) > > > > > > > > diff --git a/gdb/infrun.c b/gdb/infrun.c > > > > index 4730d290442..922d8a6913d 100644 > > > > --- a/gdb/infrun.c > > > > +++ b/gdb/infrun.c > > > > @@ -7334,8 +7334,21 @@ process_event_stop_test (struct > > > > execution_control_state *ecs) > > > > initial outermost frame, before sp was valid, would > > > > have code_addr == &_start. See the comment in > > > > frame_id::operator== > > > > for more. */ > > > > + > > > > + /* We want "nexti" to step into, not over, signal handlers > > > > invoked > > > > + by the kernel, therefore this subroutine check should not > > > > trigger > > > > + for a signal handler invocation. On most platforms, this > > > > is > > > > already > > > > + not the case, as the kernel puts a signal trampoline > > > > frame > > > > onto the > > > > + stack to handle proper return after the handler, and > > > > therefore at this > > > > + point, the current frame is a grandchild of the step > > > > frame, > > > > not a > > > > + child. However, on some platforms, the kernel actually > > > > uses > > > > a > > > > + trampoline to handle *invocation* of the handler. In > > > > that > > > > case, > > > > + when executing the first instruction of the trampoline, > > > > this > > > > check > > > > + would erroneously detect the trampoline invocation as a > > > > subroutine > > > > + call. Fix this by checking for SIGTRAMP_FRAME. */ > > > > if ((get_stack_frame_id (frame) > > > > != ecs->event_thread->control.step_stack_frame_id) > > > > + && get_frame_type (frame) != SIGTRAMP_FRAME > > > > && ((frame_unwind_caller_id (get_current_frame ()) > > > > == ecs->event_thread->control.step_stack_frame_id) > > > > && ((ecs->event_thread->control.step_stack_frame_id > > > > diff --git a/gdb/ppc-linux-tdep.c b/gdb/ppc-linux-tdep.c > > > > index 784dafa59db..7fb90799dff 100644 > > > > --- a/gdb/ppc-linux-tdep.c > > > > +++ b/gdb/ppc-linux-tdep.c > > > > @@ -83,6 +83,7 @@ > > > > #include "features/rs6000/powerpc-isa207-vsx64l.c" > > > > #include "features/rs6000/powerpc-isa207-htm-vsx64l.c" > > > > #include "features/rs6000/powerpc-e500l.c" > > > > +#include "dwarf2/frame.h" > > > > > > > > /* Shared library operations for PowerPC-Linux. */ > > > > static struct target_so_ops powerpc_so_ops; > > > > @@ -2088,6 +2089,52 @@ > > > > ppc_linux_displaced_step_prepare (gdbarch > > > > *arch, thread_info *thread, > > > > return per_inferior->disp_step_buf->prepare (thread, > > > > displaced_pc); > > > > } > > > > > > > > +/* Convert a Dwarf 2 register number to a GDB register number > > > > for > > > > Linux. */ > > > > +static int > > > > +rs6000_linux_dwarf2_reg_to_regnum (struct gdbarch *gdbarch, > > > > int > > > > num) > > > > +{ > > > > + ppc_gdbarch_tdep *tdep = > > > > gdbarch_tdep(gdbarch); > > > > + > > > > + if (0 <= num && num <= 31) > > > > + return tdep->ppc_gp0_regnum + num; > > > > + else if (32 <= num && num <= 63) > > > > + /* FIXME: jimb/2004-05-05: What should we do when the > > > > debug > > > > info > > > > + specifies registers the architecture doesn't have? Our > > > > + callers don't check the value we return. */ > > > > > > I see this comment was just copied from else where, but isn't the > > > answer > > > just: return -1 ? > > > > > > The comment about the 'return -1' at the trail of this function > > > seems > > > to > > > suggest that would be the correct thing to do. > > > > > > I guess I'm asking: do we need to add another copy of this (I > > > think > > > out > > > of date) fixme? > > > > Yes, the function is based on rs6000_dwarf2_reg_to_regnum with the > > specific changes for the Linux DWARF mappings. The comment looked > > out > > of date to me, but I left it for consistency with the other two > > functions that have the same comment. It would probably be best to > > investigate the comment further and update it in all places in a > > separate patch. > > > > Would it be acceptable to leave the comment as is for now, for > > consistency sake, and I will work on a separate cleanup patch to > > address removing the comment in the original two places and this > > additional place? I will want to look into it a bit more but I > > think > > we can just remove the comment. But I do want to verify that the > > return value is never used first. > > I took a quick look. There are three callers: 1 in dwarf2/loc.c, and > 2 > in jit.c. One of the callers in jit.c does fail to check the return > value, but this looks like a bug to me. The other callers do check > for > a -1 return value. The comment in gdbarch-gen.h is also explicit > that > the right thing to do is return -1. > > Coupled with the fact that the comment, in its current location just > makes no sense I think the best idea would be to drop it from your > new > function. Cleaning up the older code in the future would be nice, > but > is not required. But I don't think we need to introduce more > incorrect > comments just for consistency. > > Thanks, > Andrew >