Message ID | 129b33db9ac98dc806807d6595fbeec3e18edeb1.1664962429.git.aburgess@redhat.com |
---|---|
State | Committed |
Headers |
Return-Path: <gdb-patches-bounces+patchwork=sourceware.org@sourceware.org> X-Original-To: patchwork@sourceware.org Delivered-To: patchwork@sourceware.org Received: from server2.sourceware.org (localhost [IPv6:::1]) by sourceware.org (Postfix) with ESMTP id 8087E3857010 for <patchwork@sourceware.org>; Wed, 5 Oct 2022 09:35:14 +0000 (GMT) DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 8087E3857010 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=sourceware.org; s=default; t=1664962514; bh=9dMYBIrfag+/acGlcs/0Fec1XgQzg7BIcjEKI82/gB8=; h=To:Subject:Date:In-Reply-To:References:List-Id:List-Unsubscribe: List-Archive:List-Post:List-Help:List-Subscribe:From:Reply-To: From; b=YZR4FPRxmkjRAXx8fqUf2CK0QxTcvlBhvFsNzdkaqt/R2Jq4bCk5slB0Rkuxs8WtS dZGI5RRQ49PjCqtfnvzgjIcmiHreKBsttSMGG+xz0DBrHovJe0LafFTk54FGLmtftW CmUhLbhJ1Tw57wUIDT4BvkFw+x02c9hwguQJujWU= X-Original-To: gdb-patches@sourceware.org Delivered-To: gdb-patches@sourceware.org Received: from us-smtp-delivery-124.mimecast.com (us-smtp-delivery-124.mimecast.com [170.10.133.124]) by sourceware.org (Postfix) with ESMTPS id E13DB385740A for <gdb-patches@sourceware.org>; Wed, 5 Oct 2022 09:34:49 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org E13DB385740A Received: from mail-wr1-f69.google.com (mail-wr1-f69.google.com [209.85.221.69]) by relay.mimecast.com with ESMTP with STARTTLS (version=TLSv1.3, cipher=TLS_AES_128_GCM_SHA256) id us-mta-94-SRC22MUOPSacIvEPDm4sQQ-1; Wed, 05 Oct 2022 05:34:48 -0400 X-MC-Unique: SRC22MUOPSacIvEPDm4sQQ-1 Received: by mail-wr1-f69.google.com with SMTP id k30-20020adfb35e000000b0022e04708c18so3811879wrd.22 for <gdb-patches@sourceware.org>; Wed, 05 Oct 2022 02:34:48 -0700 (PDT) X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20210112; h=content-transfer-encoding:mime-version:references:in-reply-to :message-id:date:subject:cc:to:from:x-gm-message-state:from:to:cc :subject:date:message-id:reply-to; bh=9dMYBIrfag+/acGlcs/0Fec1XgQzg7BIcjEKI82/gB8=; b=O4RKcOC3oyqGUibEry5JX6qKiMGXnqj9xTCoofzDgMUtfuzjheNZNzOvP9mhLQfJEC tXJSb4knKv42Z2Xg15o+wzZJQ5LL8rPSJRK6HPReEFnT/KFYZbsxcgnOmuDCBVW0bqsp 8sQXFy38/SC739DF1ZnZ7jr+Da6ICoQKo5DEUF+5Ie+VoQANabl4PaXbvKdG7v8TP+Lt sCNHXCzrDUq/sDL8idfPbpVcajEhr+b2mnkZ8Zy6EoircD6nJw9dgt5d0oPAGWoHe1g+ 2TVHnj8ET0h9TgQzMcu0soH8zD9T7M69hPqyUyPVjh0jbMixCi+6x/RAJIsjjSbTQ+xh E6mw== X-Gm-Message-State: ACrzQf2o1Tq0lATOHuYEHyEo4PUEfad5a8DDRwv9whQMMICzfr4i83O6 Ig3przTrwcYfhfkXgi7/vXSlfTnjEBG8C7bLsMGds6u22591tIiDZAPWjCEEV2Hcq5NcSCmHKMY 9MdlgVZEFwEfRbG8kbcgYZpLPOuH5A6gV5nQ4KYgH3cTLxQ0y0J3DFNtOYiXiH08vxPjlvdsHCQ == X-Received: by 2002:a05:600c:35cf:b0:3b4:c0c2:d213 with SMTP id r15-20020a05600c35cf00b003b4c0c2d213mr2796138wmq.162.1664962486885; Wed, 05 Oct 2022 02:34:46 -0700 (PDT) X-Google-Smtp-Source: AMsMyM6s1C3BZCrd3m7C1vk1MjthMu7BnvHZ/sV09vnDDeXvmCESHJhd4x1WBDHAD22whsSvqA/WmQ== X-Received: by 2002:a05:600c:35cf:b0:3b4:c0c2:d213 with SMTP id r15-20020a05600c35cf00b003b4c0c2d213mr2796124wmq.162.1664962486563; Wed, 05 Oct 2022 02:34:46 -0700 (PDT) Received: from localhost (52.72.115.87.dyn.plus.net. [87.115.72.52]) by smtp.gmail.com with ESMTPSA id j5-20020a05600c1c0500b003b477532e66sm8884959wms.2.2022.10.05.02.34.45 (version=TLS1_3 cipher=TLS_AES_256_GCM_SHA384 bits=256/256); Wed, 05 Oct 2022 02:34:46 -0700 (PDT) To: gdb-patches@sourceware.org Subject: [PATCH 2/2] gdb: more infrun debug from breakpoint.c Date: Wed, 5 Oct 2022 10:34:40 +0100 Message-Id: <129b33db9ac98dc806807d6595fbeec3e18edeb1.1664962429.git.aburgess@redhat.com> X-Mailer: git-send-email 2.25.4 In-Reply-To: <cover.1664962429.git.aburgess@redhat.com> References: <cover.1664962429.git.aburgess@redhat.com> MIME-Version: 1.0 X-Mimecast-Spam-Score: 0 X-Mimecast-Originator: redhat.com Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset="US-ASCII"; x-default=true X-Spam-Status: No, score=-10.4 required=5.0 tests=BAYES_00, DKIMWL_WL_HIGH, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, GIT_PATCH_0, RCVD_IN_BARRACUDACENTRAL, RCVD_IN_DNSWL_NONE, SPF_HELO_NONE, SPF_NONE, 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 X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gdb-patches mailing list <gdb-patches.sourceware.org> List-Unsubscribe: <https://sourceware.org/mailman/options/gdb-patches>, <mailto:gdb-patches-request@sourceware.org?subject=unsubscribe> List-Archive: <https://sourceware.org/pipermail/gdb-patches/> List-Post: <mailto:gdb-patches@sourceware.org> List-Help: <mailto:gdb-patches-request@sourceware.org?subject=help> List-Subscribe: <https://sourceware.org/mailman/listinfo/gdb-patches>, <mailto:gdb-patches-request@sourceware.org?subject=subscribe> From: Andrew Burgess via Gdb-patches <gdb-patches@sourceware.org> Reply-To: Andrew Burgess <aburgess@redhat.com> Errors-To: gdb-patches-bounces+patchwork=sourceware.org@sourceware.org Sender: "Gdb-patches" <gdb-patches-bounces+patchwork=sourceware.org@sourceware.org> |
Series |
Some additional debug output
|
|
Commit Message
Andrew Burgess
Oct. 5, 2022, 9:34 a.m. UTC
This commit adds additional infrun debug from the breakpoint.c file. The new debug output all relates to breakpoint condition evaluation. There is already some infrun debug emitted from the breakpoint.c file, so hopefully, adding more will not be contentious. I think the functions being instrumented make sense as part of the infrun process, the inferior stops, evaluates the condition, and then either stops or continues. This new debug gives more insight into that process. I had to make the bp_location* argument to find_loc_num_by_location const, and add a declaration for find_loc_num_by_location. There should be no user visible changes unless they turn on debug output. --- gdb/breakpoint.c | 22 ++++++++++++++++++++-- 1 file changed, 20 insertions(+), 2 deletions(-)
Comments
On 2022-10-05 05:34, Andrew Burgess via Gdb-patches wrote: > This commit adds additional infrun debug from the breakpoint.c file. > The new debug output all relates to breakpoint condition evaluation. > > There is already some infrun debug emitted from the breakpoint.c file, > so hopefully, adding more will not be contentious. I think the > functions being instrumented make sense as part of the infrun process, > the inferior stops, evaluates the condition, and then either stops or > continues. This new debug gives more insight into that process. > > I had to make the bp_location* argument to find_loc_num_by_location > const, and add a declaration for find_loc_num_by_location. > > There should be no user visible changes unless they turn on debug > output. > --- > gdb/breakpoint.c | 22 ++++++++++++++++++++-- > 1 file changed, 20 insertions(+), 2 deletions(-) > > diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c > index 002f4a935b1..38546ea44ba 100644 > --- a/gdb/breakpoint.c > +++ b/gdb/breakpoint.c > @@ -165,6 +165,8 @@ static std::vector<symtab_and_line> bkpt_probe_decode_location_spec > > static bool bl_address_is_meaningful (bp_location *loc); > > +static int find_loc_num_by_location (const bp_location *loc); > + > /* update_global_location_list's modes of operation wrt to whether to > insert locations now. */ > enum ugll_insert_mode > @@ -5302,6 +5304,8 @@ bpstat_check_watchpoint (bpstat *bs) > static void > bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) > { > + INFRUN_SCOPED_DEBUG_ENTER_EXIT; > + > const struct bp_location *bl; > struct breakpoint *b; > /* Assume stop. */ > @@ -5316,6 +5320,10 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) > b = bs->breakpoint_at; > gdb_assert (b != NULL); > > + infrun_debug_printf ("thread = %s, breakpoint %d.%d", > + thread->ptid.to_string ().c_str (), > + b->number, find_loc_num_by_location (bl)); > + > /* Even if the target evaluated the condition on its end and notified GDB, we > need to do so again since GDB does not know if we stopped due to a > breakpoint or a single step breakpoint. */ > @@ -5323,6 +5331,9 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) > if (frame_id_p (b->frame_id) > && !frame_id_eq (b->frame_id, get_stack_frame_id (get_current_frame ()))) > { > + infrun_debug_printf ("incorrect frame %s not %s, not stopping", > + get_stack_frame_id (get_current_frame ()).to_string ().c_str (), > + b->frame_id.to_string ().c_str ()); > bs->stop = 0; > return; > } > @@ -5333,6 +5344,7 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) > if ((b->thread != -1 && b->thread != thread->global_num) > || (b->task != 0 && b->task != ada_get_task_number (thread))) > { > + infrun_debug_printf ("incorrect thread or task, not stopping"); > bs->stop = 0; > return; > } > @@ -5424,16 +5436,22 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) > > if (cond && !condition_result) > { > + infrun_debug_printf ("condition_result = false, not stopping"); > bs->stop = 0; > } > else if (b->ignore_count > 0) > { > + infrun_debug_printf ("ignore count %d, not stopping", > + b->ignore_count); > b->ignore_count--; > bs->stop = 0; > /* Increase the hit count even though we don't stop. */ > ++(b->hit_count); > gdb::observers::breakpoint_modified.notify (b); > - } > + } > + > + if (bs->stop) > + infrun_debug_printf ("stopping at this breakpoint"); > } I'd suggest a little change just for readability: if (cond != nullptr && !condition_result) { ... return; } if (b->ignore_count > 0) { ... return; } infrun_debug_printf ("stopping at this breakpoint"); This way, all cases where we decided not to stop are handled as early returns (there are more above in the function). And if we reach the very end of the function, it means we decided to stop. Other than that, this patch and the previous one LGTM. Simon
Simon Marchi <simark@simark.ca> writes: > On 2022-10-05 05:34, Andrew Burgess via Gdb-patches wrote: >> This commit adds additional infrun debug from the breakpoint.c file. >> The new debug output all relates to breakpoint condition evaluation. >> >> There is already some infrun debug emitted from the breakpoint.c file, >> so hopefully, adding more will not be contentious. I think the >> functions being instrumented make sense as part of the infrun process, >> the inferior stops, evaluates the condition, and then either stops or >> continues. This new debug gives more insight into that process. >> >> I had to make the bp_location* argument to find_loc_num_by_location >> const, and add a declaration for find_loc_num_by_location. >> >> There should be no user visible changes unless they turn on debug >> output. >> --- >> gdb/breakpoint.c | 22 ++++++++++++++++++++-- >> 1 file changed, 20 insertions(+), 2 deletions(-) >> >> diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c >> index 002f4a935b1..38546ea44ba 100644 >> --- a/gdb/breakpoint.c >> +++ b/gdb/breakpoint.c >> @@ -165,6 +165,8 @@ static std::vector<symtab_and_line> bkpt_probe_decode_location_spec >> >> static bool bl_address_is_meaningful (bp_location *loc); >> >> +static int find_loc_num_by_location (const bp_location *loc); >> + >> /* update_global_location_list's modes of operation wrt to whether to >> insert locations now. */ >> enum ugll_insert_mode >> @@ -5302,6 +5304,8 @@ bpstat_check_watchpoint (bpstat *bs) >> static void >> bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >> { >> + INFRUN_SCOPED_DEBUG_ENTER_EXIT; >> + >> const struct bp_location *bl; >> struct breakpoint *b; >> /* Assume stop. */ >> @@ -5316,6 +5320,10 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >> b = bs->breakpoint_at; >> gdb_assert (b != NULL); >> >> + infrun_debug_printf ("thread = %s, breakpoint %d.%d", >> + thread->ptid.to_string ().c_str (), >> + b->number, find_loc_num_by_location (bl)); >> + >> /* Even if the target evaluated the condition on its end and notified GDB, we >> need to do so again since GDB does not know if we stopped due to a >> breakpoint or a single step breakpoint. */ >> @@ -5323,6 +5331,9 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >> if (frame_id_p (b->frame_id) >> && !frame_id_eq (b->frame_id, get_stack_frame_id (get_current_frame ()))) >> { >> + infrun_debug_printf ("incorrect frame %s not %s, not stopping", >> + get_stack_frame_id (get_current_frame ()).to_string ().c_str (), >> + b->frame_id.to_string ().c_str ()); >> bs->stop = 0; >> return; >> } >> @@ -5333,6 +5344,7 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >> if ((b->thread != -1 && b->thread != thread->global_num) >> || (b->task != 0 && b->task != ada_get_task_number (thread))) >> { >> + infrun_debug_printf ("incorrect thread or task, not stopping"); >> bs->stop = 0; >> return; >> } >> @@ -5424,16 +5436,22 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >> >> if (cond && !condition_result) >> { >> + infrun_debug_printf ("condition_result = false, not stopping"); >> bs->stop = 0; >> } >> else if (b->ignore_count > 0) >> { >> + infrun_debug_printf ("ignore count %d, not stopping", >> + b->ignore_count); >> b->ignore_count--; >> bs->stop = 0; >> /* Increase the hit count even though we don't stop. */ >> ++(b->hit_count); >> gdb::observers::breakpoint_modified.notify (b); >> - } >> + } >> + >> + if (bs->stop) >> + infrun_debug_printf ("stopping at this breakpoint"); >> } > > I'd suggest a little change just for readability: > > if (cond != nullptr && !condition_result) > { > ... > return; > } > > if (b->ignore_count > 0) > { > ... > return; > } > > infrun_debug_printf ("stopping at this breakpoint"); That does look better, but unfortunately, is not correct. The problem is this line, about midway through the function: bs->stop = breakpoint_ext_lang_cond_says_stop (b); which invokes e.g. the Python Breakpoint.stop methods. It is possible that this sets the stop field back to 0. Having looked at this a little, I now have a bunch of questions about this code, like, prior to the quoted line we perform the thread and frame checks for the breakpoint, but after the quoted line we perform the condition and ignore count checks for the breakpoint. This means that for a breakpoint in the wrong thread/frame, the Python stop method is never run. But, for a breakpoint with a condition or ignore count, the stop method is run, and might choose to return True, indicating GDB should stop. We'll then check the condition, or check the ignore count, and decide that we should continue. This doesn't seem right. I'm tempted to suggest that either we should run the extension language hooks first, before any other checks, so the hook is _always_ run, but then update the docs to say that GDB might still choose to override a stop request for a breakpoint. Or, we should run the extension language hooks after all other checks, so the hook is _only_ run if GDB could stop at this breakpoint. Anyway. I've tweaked my original patches slightly and will push them, I'll follow up to this mail with what I'm actually pushing. Thanks, Andrew > > This way, all cases where we decided not to stop are handled as early > returns (there are more above in the function). And if we reach the > very end of the function, it means we decided to stop. > > Other than that, this patch and the previous one LGTM. > > Simon
Some extra debug output. --- Andrew Burgess (3): gdb: add some additional debug in mark_async_event_handler gdb: more infrun debug from breakpoint.c gdb: add missing nullptr checks in bpstat_check_breakpoint_conditions gdb/async-event.c | 6 ++++-- gdb/breakpoint.c | 30 ++++++++++++++++++++++++++---- 2 files changed, 30 insertions(+), 6 deletions(-)
Andrew Burgess <aburgess@redhat.com> writes: > Simon Marchi <simark@simark.ca> writes: > >> On 2022-10-05 05:34, Andrew Burgess via Gdb-patches wrote: >>> This commit adds additional infrun debug from the breakpoint.c file. >>> The new debug output all relates to breakpoint condition evaluation. >>> >>> There is already some infrun debug emitted from the breakpoint.c file, >>> so hopefully, adding more will not be contentious. I think the >>> functions being instrumented make sense as part of the infrun process, >>> the inferior stops, evaluates the condition, and then either stops or >>> continues. This new debug gives more insight into that process. >>> >>> I had to make the bp_location* argument to find_loc_num_by_location >>> const, and add a declaration for find_loc_num_by_location. >>> >>> There should be no user visible changes unless they turn on debug >>> output. >>> --- >>> gdb/breakpoint.c | 22 ++++++++++++++++++++-- >>> 1 file changed, 20 insertions(+), 2 deletions(-) >>> >>> diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c >>> index 002f4a935b1..38546ea44ba 100644 >>> --- a/gdb/breakpoint.c >>> +++ b/gdb/breakpoint.c >>> @@ -165,6 +165,8 @@ static std::vector<symtab_and_line> bkpt_probe_decode_location_spec >>> >>> static bool bl_address_is_meaningful (bp_location *loc); >>> >>> +static int find_loc_num_by_location (const bp_location *loc); >>> + >>> /* update_global_location_list's modes of operation wrt to whether to >>> insert locations now. */ >>> enum ugll_insert_mode >>> @@ -5302,6 +5304,8 @@ bpstat_check_watchpoint (bpstat *bs) >>> static void >>> bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >>> { >>> + INFRUN_SCOPED_DEBUG_ENTER_EXIT; >>> + >>> const struct bp_location *bl; >>> struct breakpoint *b; >>> /* Assume stop. */ >>> @@ -5316,6 +5320,10 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >>> b = bs->breakpoint_at; >>> gdb_assert (b != NULL); >>> >>> + infrun_debug_printf ("thread = %s, breakpoint %d.%d", >>> + thread->ptid.to_string ().c_str (), >>> + b->number, find_loc_num_by_location (bl)); >>> + >>> /* Even if the target evaluated the condition on its end and notified GDB, we >>> need to do so again since GDB does not know if we stopped due to a >>> breakpoint or a single step breakpoint. */ >>> @@ -5323,6 +5331,9 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >>> if (frame_id_p (b->frame_id) >>> && !frame_id_eq (b->frame_id, get_stack_frame_id (get_current_frame ()))) >>> { >>> + infrun_debug_printf ("incorrect frame %s not %s, not stopping", >>> + get_stack_frame_id (get_current_frame ()).to_string ().c_str (), >>> + b->frame_id.to_string ().c_str ()); >>> bs->stop = 0; >>> return; >>> } >>> @@ -5333,6 +5344,7 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >>> if ((b->thread != -1 && b->thread != thread->global_num) >>> || (b->task != 0 && b->task != ada_get_task_number (thread))) >>> { >>> + infrun_debug_printf ("incorrect thread or task, not stopping"); >>> bs->stop = 0; >>> return; >>> } >>> @@ -5424,16 +5436,22 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) >>> >>> if (cond && !condition_result) >>> { >>> + infrun_debug_printf ("condition_result = false, not stopping"); >>> bs->stop = 0; >>> } >>> else if (b->ignore_count > 0) >>> { >>> + infrun_debug_printf ("ignore count %d, not stopping", >>> + b->ignore_count); >>> b->ignore_count--; >>> bs->stop = 0; >>> /* Increase the hit count even though we don't stop. */ >>> ++(b->hit_count); >>> gdb::observers::breakpoint_modified.notify (b); >>> - } >>> + } >>> + >>> + if (bs->stop) >>> + infrun_debug_printf ("stopping at this breakpoint"); >>> } >> >> I'd suggest a little change just for readability: >> >> if (cond != nullptr && !condition_result) >> { >> ... >> return; >> } >> >> if (b->ignore_count > 0) >> { >> ... >> return; >> } >> >> infrun_debug_printf ("stopping at this breakpoint"); > > That does look better, but unfortunately, is not correct. The problem > is this line, about midway through the function: > > bs->stop = breakpoint_ext_lang_cond_says_stop (b); > > which invokes e.g. the Python Breakpoint.stop methods. It is possible > that this sets the stop field back to 0. > > Having looked at this a little, I now have a bunch of questions about > this code, like, prior to the quoted line we perform the thread and > frame checks for the breakpoint, but after the quoted line we perform > the condition and ignore count checks for the breakpoint. > > This means that for a breakpoint in the wrong thread/frame, the Python > stop method is never run. > > But, for a breakpoint with a condition or ignore count, the stop method > is run, and might choose to return True, indicating GDB should stop. > We'll then check the condition, or check the ignore count, and decide > that we should continue. > > This doesn't seem right. > > I'm tempted to suggest that either we should run the extension language > hooks first, before any other checks, so the hook is _always_ run, but > then update the docs to say that GDB might still choose to override a > stop request for a breakpoint. Or, we should run the extension language > hooks after all other checks, so the hook is _only_ run if GDB could > stop at this breakpoint. So, after a little more digging, it's not as bad as I thought. GDB will mostly stop a user trying to implement an extension language stop method, and add a condition to a breakpoint. So we don't need to worry about that case. The ignore count case is still open though. I think over time the ignore count implementation has drifted a little from the docs. The manual says: If a breakpoint has a positive ignore count and a condition, the condition is not checked. Once the ignore count reaches zero, @value{GDBN} resumes checking the condition. But I don't think that is true any longer. We always check the condition, and the ignore count will override the condition, but only when the condition would otherwise stop the breakpoint. Which I think makes sense? The ignore count is now ignoring the next N times the breakpoint _would_ have stopped, not just ignoring the breakpoint the next N times we hit it. Given this has been this way for a long time now, it's probably not a good idea to mess with it, and it probably makes sense for the extension language to match this behaviour. I think there's probably some documentation improvements that I could make in this area, but I don't think GDB itself needs to change. Thanks, Andrew
diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c index 002f4a935b1..38546ea44ba 100644 --- a/gdb/breakpoint.c +++ b/gdb/breakpoint.c @@ -165,6 +165,8 @@ static std::vector<symtab_and_line> bkpt_probe_decode_location_spec static bool bl_address_is_meaningful (bp_location *loc); +static int find_loc_num_by_location (const bp_location *loc); + /* update_global_location_list's modes of operation wrt to whether to insert locations now. */ enum ugll_insert_mode @@ -5302,6 +5304,8 @@ bpstat_check_watchpoint (bpstat *bs) static void bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) { + INFRUN_SCOPED_DEBUG_ENTER_EXIT; + const struct bp_location *bl; struct breakpoint *b; /* Assume stop. */ @@ -5316,6 +5320,10 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) b = bs->breakpoint_at; gdb_assert (b != NULL); + infrun_debug_printf ("thread = %s, breakpoint %d.%d", + thread->ptid.to_string ().c_str (), + b->number, find_loc_num_by_location (bl)); + /* Even if the target evaluated the condition on its end and notified GDB, we need to do so again since GDB does not know if we stopped due to a breakpoint or a single step breakpoint. */ @@ -5323,6 +5331,9 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) if (frame_id_p (b->frame_id) && !frame_id_eq (b->frame_id, get_stack_frame_id (get_current_frame ()))) { + infrun_debug_printf ("incorrect frame %s not %s, not stopping", + get_stack_frame_id (get_current_frame ()).to_string ().c_str (), + b->frame_id.to_string ().c_str ()); bs->stop = 0; return; } @@ -5333,6 +5344,7 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) if ((b->thread != -1 && b->thread != thread->global_num) || (b->task != 0 && b->task != ada_get_task_number (thread))) { + infrun_debug_printf ("incorrect thread or task, not stopping"); bs->stop = 0; return; } @@ -5424,16 +5436,22 @@ bpstat_check_breakpoint_conditions (bpstat *bs, thread_info *thread) if (cond && !condition_result) { + infrun_debug_printf ("condition_result = false, not stopping"); bs->stop = 0; } else if (b->ignore_count > 0) { + infrun_debug_printf ("ignore count %d, not stopping", + b->ignore_count); b->ignore_count--; bs->stop = 0; /* Increase the hit count even though we don't stop. */ ++(b->hit_count); gdb::observers::breakpoint_modified.notify (b); - } + } + + if (bs->stop) + infrun_debug_printf ("stopping at this breakpoint"); } /* Returns true if we need to track moribund locations of LOC's type @@ -13157,7 +13175,7 @@ enable_disable_bp_num_loc (int bp_num, int loc_num, bool enable) owner. 1-based indexing. -1 signals NOT FOUND. */ static int -find_loc_num_by_location (bp_location *loc) +find_loc_num_by_location (const bp_location *loc) { if (loc != nullptr && loc->owner != nullptr) {