From patchwork Wed Jun 27 15:30:13 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Pedro Alves X-Patchwork-Id: 28079 Received: (qmail 78011 invoked by alias); 27 Jun 2018 15:30:50 -0000 Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Unsubscribe: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org Delivered-To: mailing list gdb-patches@sourceware.org Received: (qmail 67727 invoked by uid 89); 27 Jun 2018 15:30:26 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-24.5 required=5.0 tests=AWL, BAYES_00, GIT_PATCH_0, GIT_PATCH_1, GIT_PATCH_2, GIT_PATCH_3, KAM_LAZY_DOMAIN_SECURITY, SPF_HELO_PASS autolearn=ham version=3.3.2 spammy= X-HELO: mx1.redhat.com Received: from mx3-rdu2.redhat.com (HELO mx1.redhat.com) (66.187.233.73) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Wed, 27 Jun 2018 15:30:25 +0000 Received: from smtp.corp.redhat.com (int-mx06.intmail.prod.int.rdu2.redhat.com [10.11.54.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mx1.redhat.com (Postfix) with ESMTPS id 41DDC40770F1; Wed, 27 Jun 2018 15:30:15 +0000 (UTC) Received: from [127.0.0.1] (ovpn04.gateway.prod.ext.ams2.redhat.com [10.39.146.4]) by smtp.corp.redhat.com (Postfix) with ESMTP id 7EEC12156880; Wed, 27 Jun 2018 15:30:14 +0000 (UTC) Subject: [PATCH v2] Fix Cell debugging regression (Re: [PATCH] Use thread_info and inferior pointers more throughout) To: Ulrich Weigand References: <20180627131239.5F0D5D801C0@oc3748833570.ibm.com> <393b6452-79f3-4b87-d9e2-15fbe2125dc8@redhat.com> Cc: Tom Tromey , gdb-patches@sourceware.org From: Pedro Alves Message-ID: <41f5019d-31a8-8458-1a3b-4c412fd1ddc8@redhat.com> Date: Wed, 27 Jun 2018 16:30:13 +0100 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:52.0) Gecko/20100101 Thunderbird/52.8.0 MIME-Version: 1.0 In-Reply-To: <393b6452-79f3-4b87-d9e2-15fbe2125dc8@redhat.com> On 06/27/2018 02:17 PM, Pedro Alves wrote: > On 06/27/2018 02:12 PM, Ulrich Weigand wrote: >> In the meantime I also noticed another potential issue (which is >> not related to multi-arch at all): >> >>> ps_err_e >>> ps_lgetregs (struct ps_prochandle *ph, lwpid_t lwpid, prgregset_t gregset) >>> { >>> - ptid_t ptid = ptid_build (ptid_get_pid (ph->ptid), lwpid, 0); >>> - struct regcache *regcache >>> - = get_thread_arch_regcache (ptid, target_gdbarch ()); >>> + struct regcache *regcache = get_thread_regcache (ph->thread); >> >> This change also assumes that ph->thread is the same thread >> as the one indicated by lwpid. Looking at the callers of the >> various libthread_db routines that might result in a callback >> to the ps_...regs routines, it is not immediately obvious to >> me that this is actually true. Are you sure this can never >> be called to look up registers of another thread? > > Argh, no, I completely missed that. Let me take a better look. Just OOC, I added a assertion like gdb_assert (lwpid == ph->thread->ptid.lwp ()); and it didn't trigger on x86-64, I think because libthread_db on x86-64 uses ps_get_thread_area and doesn't call into these routines. Maybe it does in some cases, like core debugging. I didn't run the whole testsuite, only a couple threaded programs. Anyways, the assertion did fail on PowerPC. Here's a new version that fixes that too. WDYT? From 66b0ff00aca764636e681392b60aacc64343ec21 Mon Sep 17 00:00:00 2001 From: Pedro Alves Date: Wed, 27 Jun 2018 12:54:50 +0100 Subject: [PATCH] Fix Cell debugging regression Commit 00431a78b28f ("Use thread_info and inferior pointers more throughout") broke Cell multi-arch debugging, because it made the proc-service routines (ps_lgetregs etc.) access registers using the SPU architecture if GDB happens to interrupt SPU code. The proc-service routines must always operate on the "main" (in this case PowerPC) architecture, because that's the register set libthread_db expects to be using. Restore the previous behavior, but wrapped in a new get_ps_regcache function with a describing comment. Also, the ps_l*regs routines have an explicit lwpid parameter that said commit missed; with the commit mentioned above, we started always reading the registers off of the current thread, which is incorrect. That is fixed by this commit too. gdb/ChangeLog: 2018-06-27 Pedro Alves * proc-service.c (get_ps_regcache): New. (ps_lgetregs, ps_lsetregs, ps_lgetfpregs) (ps_lsetfpregs): Use it. --- gdb/proc-service.c | 23 +++++++++++++++++++---- 1 file changed, 19 insertions(+), 4 deletions(-) diff --git a/gdb/proc-service.c b/gdb/proc-service.c index 04867d227b..393cd90927 100644 --- a/gdb/proc-service.c +++ b/gdb/proc-service.c @@ -126,13 +126,28 @@ ps_pdwrite (struct ps_prochandle *ph, psaddr_t addr, return ps_xfer_memory (ph, addr, (gdb_byte *) buf, size, 1); } +/* Get a regcache for LWPID using its inferior's "main" architecture, + which is the register set libthread_db expects to be using. In + multi-arch debugging scenarios, the thread's architecture may + differ from the inferior's "main" architecture. E.g., in the Cell + combined debugger, if GDB happens to interrupt SPU code, the + thread's architecture is SPU, and the main architecture is + PowerPC. */ + +static struct regcache * +get_ps_regcache (struct ps_prochandle *ph, lwpid_t lwpid) +{ + inferior *inf = ph->thread->inf; + return get_thread_arch_regcache (ptid_t (inf->pid, lwpid), inf->gdbarch); +} + /* Get the general registers of LWP LWPID within the target process PH and store them in GREGSET. */ ps_err_e ps_lgetregs (struct ps_prochandle *ph, lwpid_t lwpid, prgregset_t gregset) { - struct regcache *regcache = get_thread_regcache (ph->thread); + struct regcache *regcache = get_ps_regcache (ph, lwpid); target_fetch_registers (regcache, -1); fill_gregset (regcache, (gdb_gregset_t *) gregset, -1); @@ -146,7 +161,7 @@ ps_lgetregs (struct ps_prochandle *ph, lwpid_t lwpid, prgregset_t gregset) ps_err_e ps_lsetregs (struct ps_prochandle *ph, lwpid_t lwpid, const prgregset_t gregset) { - struct regcache *regcache = get_thread_regcache (ph->thread); + struct regcache *regcache = get_ps_regcache (ph, lwpid); supply_gregset (regcache, (const gdb_gregset_t *) gregset); target_store_registers (regcache, -1); @@ -160,7 +175,7 @@ ps_lsetregs (struct ps_prochandle *ph, lwpid_t lwpid, const prgregset_t gregset) ps_err_e ps_lgetfpregs (struct ps_prochandle *ph, lwpid_t lwpid, gdb_prfpregset_t *fpregset) { - struct regcache *regcache = get_thread_regcache (ph->thread); + struct regcache *regcache = get_ps_regcache (ph, lwpid); target_fetch_registers (regcache, -1); fill_fpregset (regcache, (gdb_fpregset_t *) fpregset, -1); @@ -175,7 +190,7 @@ ps_err_e ps_lsetfpregs (struct ps_prochandle *ph, lwpid_t lwpid, const gdb_prfpregset_t *fpregset) { - struct regcache *regcache = get_thread_regcache (ph->thread); + struct regcache *regcache = get_ps_regcache (ph, lwpid); supply_fpregset (regcache, (const gdb_fpregset_t *) fpregset); target_store_registers (regcache, -1);