From c6b486755e020095710c7494d029577ca967a13a Mon Sep 17 00:00:00 2001 From: Andrew Burgess Date: Thu, 30 Mar 2023 19:21:22 +0100 Subject: [PATCH] gdb: parse pending breakpoint thread/task immediately The initial motivation for this commit was to allow thread or inferior specific breakpoints to only be inserted within the appropriate inferior's program-space. The benefit of this is that inferiors for which the breakpoint does not apply will no longer need to stop, and then resume, for such breakpoints. This commit does not make this change, but is a refactor to allow this to happen in a later commit. The problem we currently have is that when a thread-specific (or inferior-specific) breakpoint is created, the thread (or inferior) number is only parsed by calling find_condition_and_thread_for_sals. This function is only called for non-pending breakpoints, and requires that we know the locations at which the breakpoint will be placed (for expression checking in case the breakpoint is also conditional). A consequence of this is that by the time we figure out the breakpoint is thread-specific we have already looked up locations in all program spaces. This feels wasteful -- if we knew the thread-id earlier then we could reduce the work GDB does by only looking up locations within the program space for which the breakpoint applies. Another consequence of how find_condition_and_thread_for_sals is called is that pending breakpoints don't currently know they are thread-specific, nor even that they are conditional! Additionally, by delaying parsing the thread-id, pending breakpoints can be created for non-existent threads, this is different to how non-pending breakpoints are handled, so I can do this: $ gdb -q ./gdb/testsuite/outputs/gdb.multi/pending-bp/pending-bp Reading symbols from ./gdb/testsuite/outputs/gdb.multi/pending-bp/pending-bp... (gdb) break foo thread 99 Function "foo" not defined. Make breakpoint pending on future shared library load? (y or [n]) y Breakpoint 1 (foo thread 99) pending. (gdb) r Starting program: /tmp/gdb/testsuite/outputs/gdb.multi/pending-bp/pending-bp [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib64/libthread_db.so.1". Error in re-setting breakpoint 1: Unknown thread 99. [Inferior 1 (process 3329749) exited normally] (gdb) GDB only checked the validity of 'thread 99' at the point the 'foo' location became non-pending. In contrast, if I try this: $ gdb -q ./gdb/testsuite/outputs/gdb.multi/pending-bp/pending-bp Reading symbols from ./gdb/testsuite/outputs/gdb.multi/pending-bp/pending-bp... (gdb) break main thread 99 Unknown thread 99. (gdb) GDB immediately checks if 'thread 99' exists. I think inconsistencies like this are confusing, and should be fixed if possible. In this commit the create_breakpoint function is updated so that the extra_string, which contains the thread, inferior, task, and/or condition information, is parsed immediately, even for pending breakpoints. Obviously, the condition still can't be validated until the breakpoint becomes non-pending, but the thread, inferior, and task information can be pulled from the extra-string, and can be validated early on, even for pending breakpoints. The -force-condition flag is also parsed as part of this early parsing change. There are a couple of benefits to doing this: 1. Printing of breakpoints is more consistent now. Consider creating a conditional breakpoint before this commit: (gdb) set breakpoint pending on (gdb) break pendingfunc if (0) Function "pendingfunc" not defined. Breakpoint 1 (pendingfunc if (0)) pending. (gdb) break main if (0) Breakpoint 2 at 0x401198: file /tmp/hello.c, line 18. (gdb) info breakpoints Num Type Disp Enb Address What 1 breakpoint keep y pendingfunc if (0) 2 breakpoint keep y 0x0000000000401198 in main at /tmp/hello.c:18 stop only if (0) (gdb) And after this commit: (gdb) set breakpoint pending on (gdb) break pendingfunc if (0) Function "pendingfunc" not defined. Breakpoint 1 (pendingfunc) pending. (gdb) break main if (0) Breakpoint 2 at 0x401198: file /home/andrew/tmp/hello.c, line 18. (gdb) info breakpoints Num Type Disp Enb Address What 1 breakpoint keep y pendingfunc stop only if (0) 2 breakpoint keep y 0x0000000000401198 in main at /home/andrew/tmp/hello.c:18 stop only if (0) (gdb) Notice that the display of the condition is now the same for the pending and non-pending breakpoints. The same is true for the thread, inferior, or task information in thread, inferior, or task specific breakpoints; this information is displayed on its own line rather than being part of the 'What' field. 2. We can check that the thread exists as soon as the pending breakpoint is created. Currently there is a weird difference between pending and non-pending breakpoints when creating a thread-specific breakpoint. A pending thread-specific breakpoint only checks its thread when it becomes non-pending, at which point the thread the breakpoint was intended for might have exited. Here's the behaviour before this commit: (gdb) set breakpoint pending on (gdb) break foo thread 2 Function "foo" not defined. Breakpoint 2 (foo thread 2) pending. (gdb) c Continuing. [Thread 0x7ffff7c56700 (LWP 2948835) exited] Error in re-setting breakpoint 2: Unknown thread 2. [Inferior 1 (process 2948832) exited normally] (gdb) Notice the 'Error in re-setting breakpoint 2: Unknown thread 2.' line, this was triggered when GDB tried to make the breakpoint non-pending, and GDB discovers that the thread no longer exists. Compare that to the behaviour after this commit: (gdb) set breakpoint pending on (gdb) break foo thread 2 Function "foo" not defined. Breakpoint 2 (foo) pending. (gdb) c Continuing. [Thread 0x7ffff7c56700 (LWP 2949243) exited] Thread-specific breakpoint 2 deleted - thread 2 no longer in the thread list. [Inferior 1 (process 2949240) exited normally] (gdb) Now the behaviour for pending breakpoints is identical to non-pending breakpoints, the thread specific breakpoint is removed as soon as the thread the breakpoint is associated with exits. There is an additional change; when the pending breakpoint is created prior to this patch we see this line: Breakpoint 2 (foo thread 2) pending. While after this patch we get this line: Breakpoint 2 (foo) pending. Notice that 'thread 2' has disappeared. This might look like a regression, but I don't think it is. That we said 'thread 2' before was just a consequence of the lazy parsing of the breakpoint specification, while with this patch GDB understands, and has parsed away the 'thread 2' bit of the spec. If folk think the old information was useful then this would be trivial to add back in code_breakpoint::say_where. As a result of this commit the breakpoints 'extra_string' field is now only used by bp_dprintf type breakpoints to hold the printf format and arguments. This string should always be empty for other breakpoint types. This allows some cleanup in print_breakpoint_location. In code_breakpoint::code_breakpoint I've changed an error case into an assert. This is because the error is now handled earlier in create_breakpoint. As a result we know that by this point, the extra_string will always be nullptr for anything other than a bp_dprintf style breakpoint. The find_condition_and_thread_for_sals function is now no longer needed, this was previously doing the delayed splitting of the extra string into thread, task, and condition, but this is now all done in create_breakpoint, so find_condition_and_thread_for_sals can be deleted, and the code that calls this in code_breakpoint::location_spec_to_sals can be removed. With this update this code would only ever be reached for bp_dprintf style breakpoints, and in these cases the extra_string should not contain anything other than format and args. The most interesting changes are all in create_breakpoint and in the new file break-cond-parse.c. We have a new block of code early on in create_breakpoint that is responsible for splitting the extra_string into its component parts by calling create_breakpoint_parse_arg_string a function in the new break-cond-parse.c file. This means that some of the later code can be simplified a little. The new break-cond-parse.c file implements the splitting up the extra_string and finding all the parts, as well as some self-tests for the new function. Finally, now we know all the breakpoint details, these can be stored within the breakpoint object if we end up creating a deferred breakpoint. Additionally, if we are creating a deferred bp_dprintf we can parse the extra_string to build the printf command. The implementation here aims to maintain backwards compatibility as much as possible, this means that: 1. We support abbreviations of 'thread', 'task', and 'inferior' in some places on the breakpoint line. The handling of abbreviations has (before this patch) been a little weird, so this works: (gdb) break *main th 1 And creates a breakpoint at '*main' for thread 1 only, while this does not work: (gdb) break main th 1 In this case GDB will try to find the symbol 'main th 1'. This weirdness exists before and after this patch. 2. The handling of '-force-condition' is odd, if this flag appears immediately after a condition then it will be treated as part of the condition, e.g.: (gdb) break main if 0 -force-condition No symbol "force" in current context. But we are fine with these alternatives: (gdb) break main if 0 thread 1 -force-condition (gdb) break main -force-condition if 0 Again, this is just a quirk of how the breakpoint line used to be parsed, but I've maintained this for backward compatibility. During review it was suggested that -force-condition should become an actual breakpoint flag (i.e. only valid after the 'break' command but before the function name), and I don't think that would be a terrible idea, however, that's not currently a trivial change, and I think should be done as a separate piece of work. For now, this patch just maintains the current behaviour. The implementation works by first splitting the breakpoint condition string (everything after the location specification) into a list of tokens, each token has a type and a value. (e.g. we have a THREAD token where the value is the thread-id string). The list of tokens is validated, and in some cases, tokens are merged. Then the values are extracted from the remaining token list. Consider this breakpoint command: (gdb) break main thread 1 if argc == 2 The condition string passed to create_breakpoint_parse_arg_string is going to be 'thread 1 if argc == 2', which is then split into the tokens: { THREAD: "1" } { CONDITION: "argc == 2" } The thread-id (1) and the condition string 'argc == 2' are extracted from these tokens and returns back to create_breakpoint. Now consider this breakpoint command: (gdb) break some_function if ( some_var == thread ) Here the user wants a breakpoint if 'some_var' is equal to the variable 'thread'. However, when this is initially parsed we will find these tokens: { CONDITION: "( some_var == " } { THREAD: ")" } This is a consequence of how we have to try and figure out the contents of the 'if' condition without actually parsing the expression; parsing the expression requires that we know the location in order to lookup the variables by name, and this can't be done for pending breakpoints (their location isn't known yet), and one of the points of this work is that we extract things like thread-id for pending breakpoints. And so, it is in this case that token merging takes place. We check if the value of a token appearing immediately after the CONDITION token looks valid. In this case, does ')' look like a valid thread-id. Clearly, in this case ')' does not, and so me merge the THREAD token into the condition token, giving: { CONDITION: "( some_var == thread )" } Which is what we want. I'm sure that we might still be able to come up with some edge cases where the parser makes the wrong choice. I think long term the best way to work around these would be to move the thread, inferior, task, and -force-condition flags to be "real" command options for the break command. I am looking into doing this, but can't guarantee if/when that work would be completed, so this patch should be reviewed assume that the work will never arrive (though I hope it will). Reviewed-By: Eli Zaretskii --- gdb/Makefile.in | 2 + gdb/NEWS | 4 + gdb/break-cond-parse.c | 702 ++++++++++++++++++ gdb/break-cond-parse.h | 52 ++ gdb/breakpoint.c | 372 ++-------- gdb/testsuite/gdb.ada/tasks.exp | 6 +- gdb/testsuite/gdb.base/condbreak.exp | 57 +- gdb/testsuite/gdb.base/pending.exp | 30 +- gdb/testsuite/gdb.linespec/explicit.exp | 14 +- gdb/testsuite/gdb.linespec/keywords.exp | 8 +- gdb/testsuite/gdb.mi/mi-dprintf-pending.exp | 3 +- .../gdb.multi/inferior-specific-bp.exp | 4 +- .../gdb.threads/del-pending-thread-bp-lib.c | 22 + .../gdb.threads/del-pending-thread-bp.c | 85 +++ .../gdb.threads/del-pending-thread-bp.exp | 98 +++ 15 files changed, 1152 insertions(+), 307 deletions(-) create mode 100644 gdb/break-cond-parse.c create mode 100644 gdb/break-cond-parse.h create mode 100644 gdb/testsuite/gdb.threads/del-pending-thread-bp-lib.c create mode 100644 gdb/testsuite/gdb.threads/del-pending-thread-bp.c create mode 100644 gdb/testsuite/gdb.threads/del-pending-thread-bp.exp diff --git a/gdb/Makefile.in b/gdb/Makefile.in index 13f512f5cad..bcf1ee45a70 100644 --- a/gdb/Makefile.in +++ b/gdb/Makefile.in @@ -1051,6 +1051,7 @@ COMMON_SFILES = \ break-catch-sig.c \ break-catch-syscall.c \ break-catch-throw.c \ + break-cond-parse.c \ breakpoint.c \ bt-utils.c \ btrace.c \ @@ -1320,6 +1321,7 @@ HFILES_NO_SRCDIR = \ bfd-target.h \ bfin-tdep.h \ block.h \ + break-cond-parse.h \ breakpoint.h \ bsd-kvm.h \ bsd-uthread.h \ diff --git a/gdb/NEWS b/gdb/NEWS index 19e05af625a..3ee6a61b210 100644 --- a/gdb/NEWS +++ b/gdb/NEWS @@ -10,6 +10,10 @@ function in Python. By default, the raw ptwrite payload is printed for each ptwrite that is encountered. +* For breakpoints that are created in the 'pending' state, any + 'thread' or 'task' keywords are parsed at the time the breakpoint is + created, rather than at the time the breakpoint becomes non-pending. + * Python API ** Added gdb.record.clear. Clears the trace data of the current recording. diff --git a/gdb/break-cond-parse.c b/gdb/break-cond-parse.c new file mode 100644 index 00000000000..f5fe308a923 --- /dev/null +++ b/gdb/break-cond-parse.c @@ -0,0 +1,702 @@ +/* Copyright (C) 2023 Free Software Foundation, Inc. + + This file is part of GDB. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +#include "defs.h" +#include "gdbsupport/gdb_assert.h" +#include "gdbsupport/selftest.h" +#include "test-target.h" +#include "scoped-mock-context.h" +#include "break-cond-parse.h" +#include "tid-parse.h" +#include "ada-lang.h" +#include "exceptions.h" + +/* When parsing tokens from a string, which direction are we parsing? + + Given the following string and pointer 'ptr': + + ABC DEF GHI JKL + ^ + ptr + + Parsing 'forward' will return the token 'GHI' and update 'ptr' to point + between GHI and JKL. Parsing 'backward' will return the token 'DEF' and + update 'ptr' to point between ABC and DEF. +*/ + +enum class parse_direction +{ + /* Parse the next token forwards. */ + forward, + + /* Parse the previous token backwards. */ + backward +}; + +/* Find the next token in DIRECTION from *CURR. */ + +static std::string_view +find_next_token (const char **curr, parse_direction direction) +{ + const char *tok_start, *tok_end; + + gdb_assert (**curr != '\0'); + + if (direction == parse_direction::forward) + { + *curr = skip_spaces (*curr); + tok_start = *curr; + *curr = skip_to_space (*curr); + tok_end = *curr - 1; + } + else + { + gdb_assert (direction == parse_direction::backward); + + while (isspace (**curr)) + --(*curr); + + tok_end = *curr; + + while (!isspace (**curr)) + --(*curr); + + tok_start = (*curr) + 1; + } + + return std::string_view (tok_start, tok_end - tok_start + 1); +} + +/* A class that represents a complete parsed token. Each token has a type + and a std::string_view into the original breakpoint condition string. */ + +struct token +{ + /* The types a token might take. */ + enum class type + { + /* These are the token types for the 'if', 'thread', 'inferior', and + 'task' keywords. The m_content for these token types is the value + passed to the keyword, not the keyword itself. */ + CONDITION, + THREAD, + INFERIOR, + TASK, + + /* This is the token used when we find unknown content, the m_content + for this token is the rest of the input string. */ + REST, + + /* This is the token for the -force-condition token, the m_content for + this token contains the keyword itself. */ + FORCE + }; + + token (enum type type, std::string_view content) + : m_type (type), + m_content (std::move (content)) + { + /* Nothing. */ + } + + /* Return a string representing this token. Only used for debug. */ + std::string to_string () const + { + switch (m_type) + { + case type::CONDITION: + return string_printf ("{ CONDITION: \"%s\" }", + std::string (m_content).c_str ()); + case type::THREAD: + return string_printf ("{ THREAD: \"%s\" }", + std::string (m_content).c_str ()); + case type::INFERIOR: + return string_printf ("{ INFERIOR: \"%s\" }", + std::string (m_content).c_str ()); + case type::TASK: + return string_printf ("{ TASK: \"%s\" }", + std::string (m_content).c_str ()); + case type::REST: + return string_printf ("{ REST: \"%s\" }", + std::string (m_content).c_str ()); + case type::FORCE: + return string_printf ("{ FORCE }"); + default: + return "** unknown **"; + } + } + + /* The type of this token. */ + const type &get_type () const + { + return m_type; + } + + /* Return the value of this token. */ + const std::string_view &get_value () const + { + gdb_assert (m_content.size () > 0); + return m_content; + } + + /* Extend this token with the contents of OTHER. This only makes sense + if OTHER is the next token after this one in the original string, + however, enforcing that restriction is left to the caller of this + function. + + When OTHER is a keyword/value token, e.g. 'thread 1', the m_content + for OTHER will only point to the '1'. However, as the m_content is a + std::string_view, then when we merge the m_content of OTHER into this + token we automatically merge in the 'thread' part too, as it + naturally sits between this token and OTHER. */ + + void + extend (const token &other) + { + m_content = std::string_view (this->m_content.data (), + (other.m_content.data () + - this->m_content.data () + + other.m_content.size ())); + } + +private: + /* The type of this token. */ + type m_type; + + /* The important content part of this token. The extend member function + depends on this being a std::string_view. */ + std::string_view m_content; +}; + +/* Split STR, a breakpoint condition string, into a vector of tokens where + each token represents a component of the condition. Tokens are first + parsed from the front of STR until we encounter an 'if' token. At this + point tokens are parsed from the end of STR until we encounter an + unknown token, which we assume is the other end of the 'if' condition. + If when scanning forward we encounter an unknown token then the + remainder of STR is placed into a 'rest' token (the rest of the + string), and no backward scan is performed. */ + +static std::vector +parse_all_tokens (const char *str) +{ + gdb_assert (str != nullptr); + + std::vector forward_results; + std::vector backward_results; + + const char *cond_start = nullptr; + const char *cond_end = nullptr; + parse_direction direction = parse_direction::forward; + std::vector *curr_results = &forward_results; + while (*str != '\0') + { + /* Find the next token. If moving backward and this token starts at + the same location as the condition then we must have found the + other end of the condition string -- we're done. */ + std::string_view t = find_next_token (&str, direction); + if (direction == parse_direction::backward && t.data () <= cond_start) + { + cond_end = &t.back (); + break; + } + + /* We only have a single flag option to check for. All the other + options take a value so require an additional token to be found. + Additionally, we require that this flag be at least '-f', we + don't allow it to be abbreviated to '-'. */ + if (t.length () > 1 && startswith ("-force-condition", t)) + { + curr_results->emplace_back (token::type::FORCE, t); + continue; + } + + /* Maybe the first token was the last token in the string. If this + is the case then we definitely can't try to extract a value + token. This also means that the token T is meaningless. Reset + TOK to point at the start of the unknown content and break out of + the loop. We'll record the unknown part of the string outside of + the scanning loop (below). */ + if (direction == parse_direction::forward && *str == '\0') + { + str = t.data (); + break; + } + + /* As before, find the next token and, if we are scanning backwards, + check that we have not reached the start of the condition string. */ + std::string_view v = find_next_token (&str, direction); + if (direction == parse_direction::backward && v.data () <= cond_start) + { + /* Use token T here as that must also be part of the condition + string. */ + cond_end = &t.back (); + break; + } + + /* When moving backward we will first parse the value token then the + keyword token, so swap them now. */ + if (direction == parse_direction::backward) + std::swap (t, v); + + /* Check for valid option in token T. If we find a valid option then + parse the value from the token V. Except for 'if', that's handled + differently. + + For the 'if' token we need to capture the entire condition + string, so record the start of the condition string and then + start scanning backwards looking for the end of the condition + string. + + The order of these checks is important, at least the check for + 'thread' must occur before the check for 'task'. We accept + abbreviations of these token names, and 't' should resolve to + 'thread', which will only happen if we check 'thread' first. */ + if (direction == parse_direction::forward && startswith ("if", t)) + { + cond_start = v.data (); + str = str + strlen (str); + gdb_assert (*str == '\0'); + --str; + direction = parse_direction::backward; + curr_results = &backward_results; + continue; + } + else if (startswith ("thread", t)) + curr_results->emplace_back (token::type::THREAD, v); + else if (startswith ("inferior", t)) + curr_results->emplace_back (token::type::INFERIOR, v); + else if (startswith ("task", t)) + curr_results->emplace_back (token::type::TASK, v); + else + { + /* An unknown token. If we are scanning forward then reset TOK + to point at the start of the unknown content, we record this + outside of the scanning loop (below). + + If we are scanning backward then unknown content is assumed to + be the other end of the condition string, obviously, this is + just a heuristic, we could be looking at a mistyped command + line, but this will be spotted when the condition is + eventually evaluated. + + Either way, no more scanning is required after this. */ + if (direction == parse_direction::forward) + str = t.data (); + else + { + gdb_assert (direction == parse_direction::backward); + cond_end = &v.back (); + } + break; + } + } + + if (cond_start != nullptr) + { + /* If we found the start of a condition string then we should have + switched to backward scan mode, and found the end of the condition + string. Capture the whole condition string into COND_STRING + now. */ + gdb_assert (direction == parse_direction::backward); + gdb_assert (cond_end != nullptr); + + std::string_view v (cond_start, cond_end - cond_start + 1); + + forward_results.emplace_back (token::type::CONDITION, v); + } + else if (*str != '\0') + { + /* If we didn't have a condition start pointer then we should still + be in forward scanning mode. If we didn't reach the end of the + input string (TOK is not at the null character) then the rest of + the input string is garbage that we didn't understand. + + Record the unknown content into REST. The caller of this function + will report this as an error later on. We could report the error + here, but we prefer to allow the caller to run other checks, and + prioritise other errors before reporting this problem. */ + gdb_assert (direction == parse_direction::forward); + gdb_assert (cond_end == nullptr); + + std::string_view v (str, strlen (str)); + + forward_results.emplace_back (token::type::REST, v); + } + + /* If we have tokens in the BACKWARD_RESULTS vector then this means that + we found an 'if' condition (which will be the last thing in the + FORWARD_RESULTS vector), and then we started a backward scan. + + The last tokens from the input string (those after the 'if' condition) + will be the first tokens added to the BACKWARD_RESULTS vector, so the + last items in the BACKWARD_RESULTS vector are those next to the 'if' + condition. + + Check the tokens in the BACKWARD_RESULTS vector from back to front. + If the tokens look invalid then we assume that they are actually part + of the 'if' condition, and merge the token with the 'if' condition. + If it turns out that this was incorrect and that instead the user just + messed up entering the token value, then this will show as an error + when parsing the 'if' condition. + + Doing this allows us to handle things like: + + break function if ( variable == thread ) + + Where 'thread' is a local variable within 'function'. When parsing + this we will initially see 'thread )' as a thread token with ')' as + the value. However, the following code will spot that ')' is not a + valid thread-id, and so we merge 'thread )' into the 'if' condition + string. + + This code also handles the special treatment for '-force-condition', + which exists for backwards compatibility reasons. Traditionally this + flag, if it occurred immediately after the 'if' condition, would be + treated as part of the 'if' condition. When the breakpoint condition + parsing code was rewritten, this behaviour was retained. */ + gdb_assert (backward_results.empty () + || (forward_results.back ().get_type () + == token::type::CONDITION)); + while (!backward_results.empty ()) + { + token &t = backward_results.back (); + + if (t.get_type () == token::type::FORCE) + forward_results.back ().extend (std::move (t)); + else if (t.get_type () == token::type::THREAD) + { + const char *end; + std::string v (t.get_value ()); + if (is_thread_id (v.c_str (), &end) && *end == '\0') + break; + forward_results.back ().extend (std::move (t)); + } + else if (t.get_type () == token::type::INFERIOR + || t.get_type () == token::type::TASK) + { + /* Place the token's value into a null-terminated string, parse + the string as a number and check that the entire string was + parsed. If this is true then this looks like a valid inferior + or task number, otherwise, assume an invalid id, and merge + this token with the 'if' token. */ + char *end; + std::string v (t.get_value ()); + (void) strtol (v.c_str (), &end, 0); + if (end > v.c_str () && *end == '\0') + break; + forward_results.back ().extend (std::move (t)); + } + else + gdb_assert_not_reached ("unexpected token type"); + + /* If we found an actual valid token above then we will have broken + out of the loop. We only get here if the token was merged with + the 'if' condition, in which case we can discard the last token + and then check the token before that. */ + backward_results.pop_back (); + } + + /* If after the above checks we still have some tokens in the + BACKWARD_RESULTS vector, then these need to be appended to the + FORWARD_RESULTS vector. However, we first reverse the order so that + FORWARD_RESULTS retains the tokens in the order they appeared in the + input string. */ + if (!backward_results.empty ()) + forward_results.insert (forward_results.end (), + backward_results.rbegin (), + backward_results.rend ()); + + return forward_results; +} + +/* Called when the global debug_breakpoint is true. Prints VEC to the + debug output stream. */ + +static void +dump_condition_tokens (const std::vector &vec) +{ + gdb_assert (debug_breakpoint); + + bool first = true; + std::string str = "Tokens: "; + for (const token &t : vec) + { + if (!first) + str += " "; + first = false; + str += t.to_string (); + } + breakpoint_debug_printf ("%s", str.c_str ()); +} + +/* See break-cond-parse.h. */ + +void +create_breakpoint_parse_arg_string + (const char *str, gdb::unique_xmalloc_ptr *cond_string_ptr, + int *thread_ptr, int *inferior_ptr, int *task_ptr, + gdb::unique_xmalloc_ptr *rest_ptr, bool *force_ptr) +{ + /* Set up the defaults. */ + cond_string_ptr->reset (); + rest_ptr->reset (); + *thread_ptr = -1; + *inferior_ptr = -1; + *task_ptr = -1; + *force_ptr = false; + + if (str == nullptr) + return; + + /* Split STR into a series of tokens. */ + std::vector tokens = parse_all_tokens (str); + if (debug_breakpoint) + dump_condition_tokens (tokens); + + /* Temporary variables. Initialised to the default state, then updated + as we parse TOKENS. If all of TOKENS is parsed successfully then the + state from these variables is copied into the output arguments before + the function returns. */ + int thread = -1, inferior = -1, task = -1; + bool force = false; + gdb::unique_xmalloc_ptr cond_string, rest; + + for (const token &t : tokens) + { + switch (t.get_type ()) + { + case token::type::FORCE: + force = true; + break; + case token::type::THREAD: + { + if (thread != -1) + error ("You can specify only one thread."); + if (task != -1 || inferior != -1) + error ("You can specify only one of thread, inferior, or task."); + const char *tmptok; + thread_info *thr + = parse_thread_id (std::string (t.get_value ()).c_str (), + &tmptok); + gdb_assert (*tmptok == '\0'); + thread = thr->global_num; + } + break; + case token::type::INFERIOR: + { + if (inferior != -1) + error ("You can specify only one inferior."); + if (task != -1 || thread != -1) + error ("You can specify only one of thread, inferior, or task."); + char *tmptok; + long inferior_id + = strtol (std::string (t.get_value ()).c_str (), &tmptok, 0); + if (*tmptok != '\0') + error (_("Junk '%s' after inferior keyword."), tmptok); + if (inferior_id > INT_MAX) + error (_("No inferior number '%ld'"), inferior_id); + inferior = static_cast (inferior_id); + struct inferior *inf = find_inferior_id (inferior); + if (inf == nullptr) + error (_("No inferior number '%d'"), inferior); + } + break; + case token::type::TASK: + { + if (task != -1) + error ("You can specify only one task."); + if (inferior != -1 || thread != -1) + error ("You can specify only one of thread, inferior, or task."); + char *tmptok; + long task_id + = strtol (std::string (t.get_value ()).c_str (), &tmptok, 0); + if (*tmptok != '\0') + error (_("Junk '%s' after task keyword."), tmptok); + if (task_id > INT_MAX) + error (_("Unknown task %ld"), task_id); + task = static_cast (task_id); + if (!valid_task_id (task)) + error (_("Unknown task %d."), task); + } + break; + case token::type::CONDITION: + cond_string.reset (savestring (t.get_value ().data (), + t.get_value ().size ())); + break; + case token::type::REST: + rest.reset (savestring (t.get_value ().data (), + t.get_value ().size ())); + break; + } + } + + /* Move results into the output locations. */ + *force_ptr = force; + *thread_ptr = thread; + *inferior_ptr = inferior; + *task_ptr = task; + rest_ptr->reset (rest.release ()); + cond_string_ptr->reset (cond_string.release ()); +} + +#if GDB_SELF_TEST + +namespace selftests { + +/* Run a single test of the create_breakpoint_parse_arg_string function. + INPUT is passed to create_breakpoint_parse_arg_string while all other + arguments are the expected output from + create_breakpoint_parse_arg_string. */ + +static void +test (const char *input, const char *condition, int thread = -1, + int inferior = -1, int task = -1, bool force = false, + const char *rest = nullptr, const char *error_msg = nullptr) +{ + gdb::unique_xmalloc_ptr extracted_condition; + gdb::unique_xmalloc_ptr extracted_rest; + int extracted_thread, extracted_inferior, extracted_task; + bool extracted_force_condition; + std::string exception_msg, error_str; + + if (error_msg != nullptr) + error_str = std::string (error_msg) + "\n"; + + try + { + create_breakpoint_parse_arg_string (input, &extracted_condition, + &extracted_thread, + &extracted_inferior, + &extracted_task, &extracted_rest, + &extracted_force_condition); + } + catch (const gdb_exception_error &ex) + { + string_file buf; + + exception_print (&buf, ex); + exception_msg = buf.release (); + } + + if ((condition == nullptr) != (extracted_condition.get () == nullptr) + || (condition != nullptr + && strcmp (condition, extracted_condition.get ()) != 0) + || (rest == nullptr) != (extracted_rest.get () == nullptr) + || (rest != nullptr && strcmp (rest, extracted_rest.get ()) != 0) + || thread != extracted_thread + || inferior != extracted_inferior + || task != extracted_task + || force != extracted_force_condition + || exception_msg != error_str) + { + if (run_verbose ()) + { + debug_printf ("input: '%s'\n", input); + debug_printf ("condition: '%s'\n", extracted_condition.get ()); + debug_printf ("rest: '%s'\n", extracted_rest.get ()); + debug_printf ("thread: %d\n", extracted_thread); + debug_printf ("inferior: %d\n", extracted_inferior); + debug_printf ("task: %d\n", extracted_task); + debug_printf ("forced: %s\n", + extracted_force_condition ? "true" : "false"); + debug_printf ("exception: '%s'\n", exception_msg.c_str ()); + } + + /* Report the failure. */ + SELF_CHECK (false); + } +} + +/* Wrapper for test function. Pass through the default values for all + parameters, except the last parameter, which indicates that we expect + INPUT to trigger an error. */ + +static void +test_error (const char *input, const char *error_msg) +{ + test (input, nullptr, -1, -1, -1, false, nullptr, error_msg); +} + +/* Test the create_breakpoint_parse_arg_string function. Just wraps + multiple calls to the test function above. */ + +static void +create_breakpoint_parse_arg_string_tests () +{ + gdbarch *arch = current_inferior ()->arch (); + scoped_restore_current_pspace_and_thread restore; + scoped_mock_context mock_target (arch); + + int global_thread_num = mock_target.mock_thread.global_num; + + /* Test parsing valid breakpoint condition strings. */ + test (" if blah ", "blah"); + test (" if blah thread 1", "blah", global_thread_num); + test (" if blah inferior 1", "blah", -1, 1); + test (" if blah thread 1 ", "blah", global_thread_num); + test ("thread 1 woof", nullptr, global_thread_num, -1, -1, false, "woof"); + test ("thread 1 X", nullptr, global_thread_num, -1, -1, false, "X"); + test (" if blah thread 1 -force-condition", "blah", global_thread_num, + -1, -1, true); + test (" -force-condition if blah thread 1", "blah", global_thread_num, + -1, -1, true); + test (" -force-condition if blah thread 1 ", "blah", global_thread_num, + -1, -1, true); + test ("thread 1 -force-condition if blah", "blah", global_thread_num, + -1, -1, true); + test ("if (A::outer::func ())", "(A::outer::func ())"); + test ("if ( foo == thread )", "( foo == thread )"); + test ("if ( foo == thread ) inferior 1", "( foo == thread )", -1, 1); + test ("if ( foo == thread ) thread 1", "( foo == thread )", + global_thread_num); + test ("if foo == thread", "foo == thread"); + test ("if foo == thread 1", "foo ==", global_thread_num); + + /* Test parsing some invalid breakpoint condition strings. */ + test_error ("thread 1 if foo == 123 thread 1", + "You can specify only one thread."); + test_error ("thread 1 if foo == 123 inferior 1", + "You can specify only one of thread, inferior, or task."); + test_error ("thread 1 if foo == 123 task 1", + "You can specify only one of thread, inferior, or task."); + test_error ("inferior 1 if foo == 123 inferior 1", + "You can specify only one inferior."); + test_error ("inferior 1 if foo == 123 thread 1", + "You can specify only one of thread, inferior, or task."); + test_error ("inferior 1 if foo == 123 task 1", + "You can specify only one of thread, inferior, or task."); + test_error ("thread 1.2.3", "Invalid thread ID: 1.2.3"); + test_error ("thread 1/2", "Invalid thread ID: 1/2"); + test_error ("thread 1xxx", "Invalid thread ID: 1xxx"); + test_error ("inferior 1xxx", "Junk 'xxx' after inferior keyword."); + test_error ("task 1xxx", "Junk 'xxx' after task keyword."); +} + +} // namespace selftests +#endif /* GDB_SELF_TEST */ + +void _initialize_break_cond_parse (); +void +_initialize_break_cond_parse () +{ +#if GDB_SELF_TEST + selftests::register_test + ("create_breakpoint_parse_arg_string", + selftests::create_breakpoint_parse_arg_string_tests); +#endif +} diff --git a/gdb/break-cond-parse.h b/gdb/break-cond-parse.h new file mode 100644 index 00000000000..cbee70f4e9e --- /dev/null +++ b/gdb/break-cond-parse.h @@ -0,0 +1,52 @@ +/* Copyright (C) 2023 Free Software Foundation, Inc. + + This file is part of GDB. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +#if !defined (BREAK_COND_PARSE_H) +#define BREAK_COND_PARSE_H 1 + +/* Given TOK, a string possibly containing a condition, thread, inferior, + task and force-condition flag, as accepted by the 'break' command, + extract the condition string, thread, inferior, task number, and the + force_condition flag, then set *COND_STRING, *THREAD, *INFERIOR, *TASK, + and *FORCE. + + As TOK is parsed, if an unknown keyword is encountered before the 'if' + keyword then everything starting from the unknown keyword is placed into + *REST. + + Both *COND and *REST are initialized to nullptr. If no 'if' keyword is + found then *COND will be returned as nullptr. If no unknown content is + found then *REST is returned as nullptr. + + If no thread is found, *THREAD is set to -1. If no inferior is found, + *INFERIOR is set to -1. If no task is found, *TASK is set to -1. If + the -force-condition flag is not found then *FORCE is set to false. + + Due to the free-form nature that the string TOK might take (a 'thread' + keyword can appear before or after an 'if' condition) then we end up + having to check for keywords from both the start of TOK and the end of + TOK. + + If TOK is nullptr, or TOK is the empty string, then the output variables + are all given their default values. */ + +extern void create_breakpoint_parse_arg_string + (const char *tok, gdb::unique_xmalloc_ptr *cond_string, + int *thread, int *inferior, int *task, + gdb::unique_xmalloc_ptr *rest, bool *force); + +#endif diff --git a/gdb/breakpoint.c b/gdb/breakpoint.c index eab3b937c13..9e30dbc1a93 100644 --- a/gdb/breakpoint.c +++ b/gdb/breakpoint.c @@ -70,6 +70,7 @@ #include "cli/cli-style.h" #include "cli/cli-decode.h" #include +#include "break-cond-parse.h" /* readline include files */ #include "readline/tilde.h" @@ -6321,20 +6322,7 @@ print_breakpoint_location (const breakpoint *b, const bp_location *loc) uiout->field_stream ("at", stb); } else - { - uiout->field_string ("pending", b->locspec->to_string ()); - /* If extra_string is available, it could be holding a condition - or dprintf arguments. In either case, make sure it is printed, - too, but only for non-MI streams. */ - if (!uiout->is_mi_like_p () && b->extra_string != NULL) - { - if (b->type == bp_dprintf) - uiout->text (","); - else - uiout->text (" "); - uiout->text (b->extra_string.get ()); - } - } + uiout->field_string ("pending", b->locspec->to_string ()); if (loc && is_breakpoint (b) && breakpoint_condition_evaluation_mode () == condition_evaluation_target @@ -8749,8 +8737,8 @@ code_breakpoint::code_breakpoint (struct gdbarch *gdbarch_, command line, otherwise it's an error. */ if (type == bp_dprintf) update_dprintf_command_list (this); - else if (extra_string != nullptr) - error (_("Garbage '%s' at end of command"), extra_string.get ()); + else + gdb_assert (extra_string == nullptr); /* The order of the locations is now stable. Set the location condition using the location's number. */ @@ -8978,197 +8966,6 @@ check_fast_tracepoint_sals (struct gdbarch *gdbarch, } } -/* Given TOK, a string specification of condition and thread, as accepted - by the 'break' command, extract the condition string into *COND_STRING. - If no condition string is found then *COND_STRING is set to nullptr. - - If the breakpoint specification has an associated thread, task, or - inferior, these are extracted into *THREAD, *TASK, and *INFERIOR - respectively, otherwise these arguments are set to -1 (for THREAD and - INFERIOR) or 0 (for TASK). - - PC identifies the context at which the condition should be parsed. */ - -static void -find_condition_and_thread (const char *tok, CORE_ADDR pc, - gdb::unique_xmalloc_ptr *cond_string, - int *thread, int *inferior, int *task, - gdb::unique_xmalloc_ptr *rest) -{ - cond_string->reset (); - *thread = -1; - *inferior = -1; - *task = -1; - rest->reset (); - bool force = false; - - while (tok && *tok) - { - const char *end_tok; - int toklen; - const char *cond_start = NULL; - const char *cond_end = NULL; - - tok = skip_spaces (tok); - - if ((*tok == '"' || *tok == ',') && rest) - { - rest->reset (savestring (tok, strlen (tok))); - break; - } - - end_tok = skip_to_space (tok); - - toklen = end_tok - tok; - - if (toklen >= 1 && strncmp (tok, "if", toklen) == 0) - { - tok = cond_start = end_tok + 1; - try - { - parse_exp_1 (&tok, pc, block_for_pc (pc), 0); - } - catch (const gdb_exception_error &) - { - if (!force) - throw; - else - tok = tok + strlen (tok); - } - cond_end = tok; - cond_string->reset (savestring (cond_start, cond_end - cond_start)); - } - else if (toklen >= 1 && strncmp (tok, "-force-condition", toklen) == 0) - { - tok = tok + toklen; - force = true; - } - else if (toklen >= 1 && strncmp (tok, "thread", toklen) == 0) - { - const char *tmptok; - struct thread_info *thr; - - if (*thread != -1) - error(_("You can specify only one thread.")); - - if (*task != -1) - error (_("You can specify only one of thread or task.")); - - if (*inferior != -1) - error (_("You can specify only one of inferior or thread.")); - - tok = end_tok + 1; - thr = parse_thread_id (tok, &tmptok); - if (tok == tmptok) - error (_("Junk after thread keyword.")); - *thread = thr->global_num; - tok = tmptok; - } - else if (toklen >= 1 && strncmp (tok, "inferior", toklen) == 0) - { - if (*inferior != -1) - error(_("You can specify only one inferior.")); - - if (*task != -1) - error (_("You can specify only one of inferior or task.")); - - if (*thread != -1) - error (_("You can specify only one of inferior or thread.")); - - char *tmptok; - tok = end_tok + 1; - *inferior = strtol (tok, &tmptok, 0); - if (tok == tmptok) - error (_("Junk after inferior keyword.")); - if (!valid_global_inferior_id (*inferior)) - error (_("Unknown inferior number %d."), *inferior); - tok = tmptok; - } - else if (toklen >= 1 && strncmp (tok, "task", toklen) == 0) - { - char *tmptok; - - if (*task != -1) - error(_("You can specify only one task.")); - - if (*thread != -1) - error (_("You can specify only one of thread or task.")); - - if (*inferior != -1) - error (_("You can specify only one of inferior or task.")); - - tok = end_tok + 1; - *task = strtol (tok, &tmptok, 0); - if (tok == tmptok) - error (_("Junk after task keyword.")); - if (!valid_task_id (*task)) - error (_("Unknown task %d."), *task); - tok = tmptok; - } - else if (rest) - { - rest->reset (savestring (tok, strlen (tok))); - break; - } - else - error (_("Junk at end of arguments.")); - } -} - -/* Call 'find_condition_and_thread' for each sal in SALS until a parse - succeeds. The parsed values are written to COND_STRING, THREAD, - TASK, and REST. See the comment of 'find_condition_and_thread' - for the description of these parameters and INPUT. */ - -static void -find_condition_and_thread_for_sals (const std::vector &sals, - const char *input, - gdb::unique_xmalloc_ptr *cond_string, - int *thread, int *inferior, int *task, - gdb::unique_xmalloc_ptr *rest) -{ - int num_failures = 0; - for (auto &sal : sals) - { - gdb::unique_xmalloc_ptr cond; - int thread_id = -1; - int inferior_id = -1; - int task_id = -1; - gdb::unique_xmalloc_ptr remaining; - - /* Here we want to parse 'arg' to separate condition from thread - number. But because parsing happens in a context and the - contexts of sals might be different, try each until there is - success. Finding one successful parse is sufficient for our - goal. When setting the breakpoint we'll re-parse the - condition in the context of each sal. */ - try - { - find_condition_and_thread (input, sal.pc, &cond, &thread_id, - &inferior_id, &task_id, &remaining); - *cond_string = std::move (cond); - /* A value of -1 indicates that these fields are unset. At most - one of these fields should be set (to a value other than -1) - at this point. */ - gdb_assert (((thread_id == -1 ? 1 : 0) - + (task_id == -1 ? 1 : 0) - + (inferior_id == -1 ? 1 : 0)) >= 2); - *thread = thread_id; - *inferior = inferior_id; - *task = task_id; - *rest = std::move (remaining); - break; - } - catch (const gdb_exception_error &e) - { - num_failures++; - /* If no sal remains, do not continue. */ - if (num_failures == sals.size ()) - throw; - } - } -} - /* Decode a static tracepoint marker spec. */ static std::vector @@ -9285,6 +9082,46 @@ create_breakpoint (struct gdbarch *gdbarch, ? (extra_string != nullptr && !parse_extra) : (extra_string == nullptr || parse_extra)); + /* Will hold either copies of the similarly named function argument, or + will hold a modified version of the function argument, depending on + the value of PARSE_EXTRA. */ + gdb::unique_xmalloc_ptr cond_string_copy; + gdb::unique_xmalloc_ptr extra_string_copy; + + if (parse_extra) + { + /* Parse EXTRA_STRING splitting the parts out. */ + create_breakpoint_parse_arg_string (extra_string, &cond_string_copy, + &thread, &inferior, &task, + &extra_string_copy, + &force_condition); + + /* We could check that EXTRA_STRING_COPY is empty at this point -- it + should be, as we only get here for things that are not bp_dprintf, + however, we prefer to give the location spec parser a chance to + run first, this means the user will get errors about invalid + location spec instead of an error about garbage at the end of the + command line. + + We still do the EXTRA_STRING_COPY is empty check, just later in + this function. */ + + gdb_assert (thread == -1 || thread > 0); + gdb_assert (task == -1 || task > 0); + gdb_assert (inferior == -1 || inferior > 0); + } + else + { + if (cond_string != nullptr) + cond_string_copy.reset (xstrdup (cond_string)); + if (extra_string != nullptr) + extra_string_copy.reset (xstrdup (extra_string)); + } + + /* Clear these. Updated values are now held in the *_copy locals. */ + cond_string = nullptr; + extra_string = nullptr; + try { ops->create_sals_from_location_spec (locspec, &canonical); @@ -9320,6 +9157,13 @@ create_breakpoint (struct gdbarch *gdbarch, throw; } + /* Only bp_dprintf breakpoints should have anything in EXTRA_STRING_COPY + by this point. For all other breakpoints this indicates an error. We + could place this check earlier in the function, but we prefer to see + errors from the location spec parser before we see this error message. */ + if (type_wanted != bp_dprintf && extra_string_copy.get () != nullptr) + error (_("Garbage '%s' at end of command"), extra_string_copy.get ()); + if (!pending && canonical.lsals.empty ()) return 0; @@ -9343,63 +9187,31 @@ create_breakpoint (struct gdbarch *gdbarch, breakpoint. */ if (!pending) { - gdb::unique_xmalloc_ptr cond_string_copy; - gdb::unique_xmalloc_ptr extra_string_copy; - - if (parse_extra) + /* Check the validity of the condition. We should error out if the + condition is invalid at all of the locations and if it is not + forced. In the PARSE_EXTRA case above, this check is done when + parsing the EXTRA_STRING. */ + if (cond_string_copy.get () != nullptr && !force_condition) { - gdb_assert (type_wanted != bp_dprintf); - - gdb::unique_xmalloc_ptr rest; - gdb::unique_xmalloc_ptr cond; - + int num_failures = 0; const linespec_sals &lsal = canonical.lsals[0]; - - find_condition_and_thread_for_sals (lsal.sals, extra_string, - &cond, &thread, &inferior, - &task, &rest); - - if (rest.get () != nullptr && *(rest.get ()) != '\0') - error (_("Garbage '%s' at end of command"), rest.get ()); - - cond_string_copy = std::move (cond); - extra_string_copy = std::move (rest); - } - else - { - /* Check the validity of the condition. We should error out - if the condition is invalid at all of the locations and - if it is not forced. In the PARSE_EXTRA case above, this - check is done when parsing the EXTRA_STRING. */ - if (cond_string != nullptr && !force_condition) + for (const auto &sal : lsal.sals) { - int num_failures = 0; - const linespec_sals &lsal = canonical.lsals[0]; - for (const auto &sal : lsal.sals) + const char *cond = cond_string_copy.get (); + try { - const char *cond = cond_string; - try - { - parse_exp_1 (&cond, sal.pc, block_for_pc (sal.pc), 0); - /* One success is sufficient to keep going. */ - break; - } - catch (const gdb_exception_error &) - { - num_failures++; - /* If this is the last sal, error out. */ - if (num_failures == lsal.sals.size ()) - throw; - } + parse_exp_1 (&cond, sal.pc, block_for_pc (sal.pc), 0); + /* One success is sufficient to keep going. */ + break; + } + catch (const gdb_exception_error &) + { + num_failures++; + /* If this is the last sal, error out. */ + if (num_failures == lsal.sals.size ()) + throw; } } - - /* Create a private copy of condition string. */ - if (cond_string) - cond_string_copy.reset (xstrdup (cond_string)); - /* Create a private copy of any extra string. */ - if (extra_string) - extra_string_copy.reset (xstrdup (extra_string)); } ops->create_breakpoints_sal (gdbarch, &canonical, @@ -9416,21 +9228,16 @@ create_breakpoint (struct gdbarch *gdbarch, type_wanted); b->locspec = locspec->clone (); - if (parse_extra) - b->cond_string = NULL; - else - { - /* Create a private copy of condition string. */ - b->cond_string.reset (cond_string != NULL - ? xstrdup (cond_string) - : NULL); - b->thread = thread; - } + /* Create a private copy of the condition string. */ + b->cond_string = std::move (cond_string_copy); + + b->thread = thread; + b->task = task; + b->inferior = inferior; /* Create a private copy of any extra string. */ - b->extra_string.reset (extra_string != NULL - ? xstrdup (extra_string) - : NULL); + b->extra_string = std::move (extra_string_copy); + b->ignore_count = ignore_count; b->disposition = tempflag ? disp_del : disp_donttouch; b->condition_not_parsed = 1; @@ -9439,9 +9246,12 @@ create_breakpoint (struct gdbarch *gdbarch, && type_wanted != bp_hardware_breakpoint) || thread != -1) b->pspace = current_program_space; + if (b->type == bp_dprintf) + update_dprintf_command_list (b.get ()); + install_breakpoint (internal, std::move (b), 0); } - + if (canonical.lsals.size () > 1) { warning (_("Multiple breakpoints were set.\nUse the " @@ -13180,24 +12990,6 @@ code_breakpoint::location_spec_to_sals (location_spec *locspec, { for (auto &sal : sals) resolve_sal_pc (&sal); - if (condition_not_parsed && extra_string != NULL) - { - gdb::unique_xmalloc_ptr local_cond, local_extra; - int local_thread, local_task, local_inferior; - - find_condition_and_thread_for_sals (sals, extra_string.get (), - &local_cond, &local_thread, - &local_inferior, - &local_task, &local_extra); - gdb_assert (cond_string == nullptr); - if (local_cond != nullptr) - cond_string = std::move (local_cond); - thread = local_thread; - task = local_task; - if (local_extra != nullptr) - extra_string = std::move (local_extra); - condition_not_parsed = 0; - } if (type == bp_static_tracepoint) { diff --git a/gdb/testsuite/gdb.ada/tasks.exp b/gdb/testsuite/gdb.ada/tasks.exp index 6e588408c9d..ecbfbd5522f 100644 --- a/gdb/testsuite/gdb.ada/tasks.exp +++ b/gdb/testsuite/gdb.ada/tasks.exp @@ -55,11 +55,11 @@ gdb_test "watch j task 1 task 3" "You can specify only one task\\." # Check that attempting to combine 'task' and 'thread' gives an error. gdb_test "break break_me task 1 thread 1" \ - "You can specify only one of thread or task\\." + "You can specify only one of thread, inferior, or task\\." gdb_test "break break_me thread 1 task 1" \ - "You can specify only one of thread or task\\." + "You can specify only one of thread, inferior, or task\\." gdb_test "break break_me inferior 1 task 1" \ - "You can specify only one of inferior or task\\." + "You can specify only one of thread, inferior, or task\\." gdb_test "watch j task 1 thread 1" \ "You can specify only one of thread or task\\." gdb_test "watch j thread 1 task 1" \ diff --git a/gdb/testsuite/gdb.base/condbreak.exp b/gdb/testsuite/gdb.base/condbreak.exp index 65d19b3e7a9..3b619e47bfc 100644 --- a/gdb/testsuite/gdb.base/condbreak.exp +++ b/gdb/testsuite/gdb.base/condbreak.exp @@ -179,6 +179,10 @@ gdb_test "break -q main if (1==1) thread 999" \ "Unknown thread 999\\." gdb_test "break -q main thread 999 if (1==1)" \ "Unknown thread 999\\." +gdb_test "break -q main if (1==1) thread 999 -force-condition" \ + "Unknown thread 999\\." +gdb_test "break -q main thread 999 if (1==1) -force-condition" \ + "Unknown thread 999\\." # Verify that both if and thread can be distinguished from a breakpoint # address expression. @@ -186,20 +190,71 @@ gdb_test "break *main if (1==1) thread 999" \ "Unknown thread 999\\." gdb_test "break *main thread 999 if (1==1)" \ "Unknown thread 999\\." +gdb_test "break *main if (1==1) thread 999 -force-condition" \ + "Unknown thread 999\\." +gdb_test "break *main thread 999 if (1==1) -force-condition" \ + "Unknown thread 999\\." # Similarly for task. gdb_test "break *main if (1==1) task 999" \ "Unknown task 999\\." gdb_test "break *main task 999 if (1==1)" \ "Unknown task 999\\." +gdb_test "break *main if (1==1) task 999 -force-condition" \ + "Unknown task 999\\." +gdb_test "break *main task 999 if (1==1) -force-condition" \ + "Unknown task 999\\." -# GDB accepts abbreviations for "thread" and "task". +# GDB accepts abbreviations for "thread", "task" and +# "-force-condition", when these keywords appear after +# the breakpoint condition. gdb_test "break *main if (1==1) t 999" \ "Unknown thread 999\\." gdb_test "break *main if (1==1) th 999" \ "Unknown thread 999\\." gdb_test "break *main if (1==1) ta 999" \ "Unknown task 999\\." +gdb_test "break *main if (1==1) t 999 -force" \ + "Unknown thread 999\\." +gdb_test "break *main if (1==1) th 999 -force" \ + "Unknown thread 999\\." +gdb_test "break *main if (1==1) ta 999 -force" \ + "Unknown task 999\\." + +# Check the use of abbreviations before the condition. This works +# because, when the location spec starts with '*' GDB is able to +# figure out that the complete location is '*main'. +gdb_test "break *main t 999 if (1==1)" \ + "Unknown thread 999\\." +gdb_test "break *main th 999 if (1==1)" \ + "Unknown thread 999\\." +gdb_test "break *main ta 999 if (1==1)" \ + "Unknown task 999\\." +gdb_test "break *main t 999 -force if (1==1)" \ + "Unknown thread 999\\." +gdb_test "break *main th 999 -force if (1==1)" \ + "Unknown thread 999\\." +gdb_test "break *main ta 999 -force if (1==1)" \ + "Unknown task 999\\." + +# However, when the location spec doesn't start with '*' GDB relies on +# the linespec parser to spot the keyword which marks the end of the +# linespec, and this parser doesn't check for abbreviations. +gdb_test "with breakpoint pending off -- break main t 999 if (1==1)" \ + "Function \"main t 999\" not defined\\." +gdb_test "with breakpoint pending off -- break main th 999 if (1==1)" \ + "Function \"main th 999\" not defined\\." +gdb_test "with breakpoint pending off -- break main ta 999 if (1==1)" \ + "Function \"main ta 999\" not defined\\." + +# GDB does not treat a "-force-condition" flag that appears +# immediately after the condition as the flag, but instead treats it +# as " - force - condition", that is, subtraction of the symbol +# "force" followed by subtraction of symbol "context". This is really +# just a quirk of how this used to be implemented, and should maybe be +# changed in the future. However, for now GDB retains this behaviour. +gdb_test "break *main if (1==1) -force-condition" \ + "No symbol \"force\" in current context\\." set test "run until breakpoint at marker3" gdb_test_multiple "continue" $test { diff --git a/gdb/testsuite/gdb.base/pending.exp b/gdb/testsuite/gdb.base/pending.exp index 5ee31d32389..833e084db82 100644 --- a/gdb/testsuite/gdb.base/pending.exp +++ b/gdb/testsuite/gdb.base/pending.exp @@ -170,7 +170,8 @@ gdb_test "info break" \ \[\t \]+stop only if k == 1.* \[\t \]+print k.* \[0-9\]+\[\t \]+breakpoint keep y.* in main at .*$srcfile:$mainline.* -\[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*pendshr.c:$bp2_loc if x > 3.*" \ +\[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*pendshr.c:$bp2_loc.* +\\s+stop only if x > 3.*" \ "multiple pending breakpoints" @@ -195,8 +196,10 @@ gdb_test "info break" \ \[\t \]+stop only if k == 1.* \[\t \]+print k.* \[0-9\]+\[\t \]+breakpoint keep y.* in main at .*$srcfile:$mainline.* -\[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*pendshr.c:$bp2_loc if x > 3.* -\[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*pendshr.c:$bp3_loc.*ignore next 2 hits.*" \ +\[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*pendshr.c:$bp2_loc.* +\\s+stop only if x > 3.* +\[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*pendshr.c:$bp3_loc.* +\\s+ignore next 2 hits.*" \ "multiple pending breakpoints 2" # @@ -267,3 +270,24 @@ gdb_test "info break" \ \[0-9\]+\[\t \]+breakpoint keep y.* in main at .*$srcfile:$mainline.* \[0-9\]+\[\t \]+breakpoint keep y.*PENDING.*imaginary.*" \ "verify pending breakpoint after restart" + +# Test GDB's parsing of pending breakpoint thread and condition. + +gdb_test_no_output "set breakpoint pending on" +gdb_test "break foo if (unknown_var && another_unknown_var) thread 1" \ + "Breakpoint $decimal \\(foo\\) pending\\." +set bpnum [get_integer_valueof "\$bpnum" "*INVALID" \ + "get number for foo breakpoint"] + +if {[gdb_protocol_is_remote]} { + set evals_re "(?: \\(\[^) \]+ evals\\))?" +} else { + set evals_re "" +} + +gdb_test "info breakpoints $bpnum" \ + [multi_line \ + "$bpnum\\s+breakpoint\\s+keep\\s+y\\s+\\s+foo" \ + "\\s+stop only if \\(unknown_var && another_unknown_var\\)${evals_re}" \ + "\\s+stop only in thread 1"] \ + "check pending breakpoint on foo" diff --git a/gdb/testsuite/gdb.linespec/explicit.exp b/gdb/testsuite/gdb.linespec/explicit.exp index 60183e98e1e..e8ae10a27df 100644 --- a/gdb/testsuite/gdb.linespec/explicit.exp +++ b/gdb/testsuite/gdb.linespec/explicit.exp @@ -575,22 +575,30 @@ namespace eval $testfile { allow-pending]} { fail "set $tst" } else { - gdb_test "info break" ".*PENDING.*myfunction if foofoofoo == 1.*" $tst + gdb_test "info break" ".*PENDING.*myfunction\r\n\\s+stop only if foofoofoo == 1.*" $tst } gdb_exit gdb_start + if {[target_info gdb_protocol] == "extended-remote"} { + set evals_re "(?: \\(\[^) \]+ evals\\))?" + } else { + set evals_re "" + } + set tst "pending valid conditional explicit breakpoint" if {![gdb_breakpoint "-func myfunction if arg == 0" \ allow-pending]} { fail "set $tst" } else { - gdb_test "info break" ".*PENDING.*myfunction if arg == 0" $tst + gdb_test "info break" \ + ".*PENDING.*myfunction\r\n\\s+stop only if arg == 0${evals_re}" \ + $tst gdb_load [standard_output_file $exefile] gdb_test "info break" \ - ".*in myfunction at .*$srcfile:.*stop only if arg == 0.*" \ + ".*in myfunction at .*$srcfile:.*stop only if arg == 0${evals_re}" \ "$tst resolved" } diff --git a/gdb/testsuite/gdb.linespec/keywords.exp b/gdb/testsuite/gdb.linespec/keywords.exp index 36a919c8be2..d2596d2b357 100644 --- a/gdb/testsuite/gdb.linespec/keywords.exp +++ b/gdb/testsuite/gdb.linespec/keywords.exp @@ -55,7 +55,7 @@ with_test_prefix "trailing whitespace" { gdb_test "break thread 123" "Unknown thread 123\\." gdb_test "break thread foo" "Invalid thread ID: foo" gdb_test "break task 123" "Unknown task 123\\." -gdb_test "break task foo" "Junk after task keyword\\." +gdb_test "break task foo" "Junk 'foo' after task keyword\\." gdb_breakpoint "thread if 0" "message" # These are also NULL locations, but using a subsequent keyword @@ -63,9 +63,9 @@ gdb_breakpoint "thread if 0" "message" gdb_test "break thread thread" "Invalid thread ID: thread" gdb_test "break thread task" "Invalid thread ID: task" gdb_test "break thread if" "Invalid thread ID: if" -gdb_test "break task task" "Junk after task keyword\\." -gdb_test "break task thread" "Junk after task keyword\\." -gdb_test "break task if" "Junk after task keyword\\." +gdb_test "break task task" "Junk 'task' after task keyword\\." +gdb_test "break task thread" "Junk 'thread' after task keyword\\." +gdb_test "break task if" "Junk 'if' after task keyword\\." # Test locations containing keyword followed by keyword. gdb_test "break thread thread 123" "Unknown thread 123\\." diff --git a/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp b/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp index fd5684bd2b1..4cf6dec4b41 100644 --- a/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp +++ b/gdb/testsuite/gdb.mi/mi-dprintf-pending.exp @@ -50,7 +50,8 @@ set bp_location1 [gdb_get_line_number "set breakpoint 1 here"] # Set pending dprintf via MI. set bp [mi_make_breakpoint_pending -number "1" -type "dprintf" \ -disp "keep" -enabled "y" -pending "pendfunc1" \ - -original-location "pendfunc1"] + -original-location "pendfunc1" \ + -script {\["printf \\\"hello\\\""\]}] mi_gdb_test "-dprintf-insert -f pendfunc1 \"hello\"" \ ".*\\^done,$bp" "mi set dprintf" diff --git a/gdb/testsuite/gdb.multi/inferior-specific-bp.exp b/gdb/testsuite/gdb.multi/inferior-specific-bp.exp index 5cc451b0ecc..46efe6f54bc 100644 --- a/gdb/testsuite/gdb.multi/inferior-specific-bp.exp +++ b/gdb/testsuite/gdb.multi/inferior-specific-bp.exp @@ -51,9 +51,9 @@ if {![runto_main]} { # this should fail. Try with the keywords in both orders just in case the # parser has a bug. gdb_test "break foo thread 1.1 inferior 1" \ - "You can specify only one of inferior or thread\\." + "You can specify only one of thread, inferior, or task\\." gdb_test "break foo inferior 1 thread 1.1" \ - "You can specify only one of inferior or thread\\." + "You can specify only one of thread, inferior, or task\\." # Try to create a breakpoint using the 'inferior' keyword multiple times. gdb_test "break foo inferior 1 inferior 2" \ diff --git a/gdb/testsuite/gdb.threads/del-pending-thread-bp-lib.c b/gdb/testsuite/gdb.threads/del-pending-thread-bp-lib.c new file mode 100644 index 00000000000..15d1b9833dd --- /dev/null +++ b/gdb/testsuite/gdb.threads/del-pending-thread-bp-lib.c @@ -0,0 +1,22 @@ +/* Copyright 2023 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +int global_var = 0; + +void +foo (int arg) +{ + global_var = arg; +} diff --git a/gdb/testsuite/gdb.threads/del-pending-thread-bp.c b/gdb/testsuite/gdb.threads/del-pending-thread-bp.c new file mode 100644 index 00000000000..6fc76dbf08c --- /dev/null +++ b/gdb/testsuite/gdb.threads/del-pending-thread-bp.c @@ -0,0 +1,85 @@ +/* Copyright 2023 Free Software Foundation, Inc. + + This program is free software; you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation; either version 3 of the License, or + (at your option) any later version. + + This program is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with this program. If not, see . */ + +#include +#include +#include + +pthread_barrier_t barrier; + +static void +barrier_wait (pthread_barrier_t *b) +{ + int res = pthread_barrier_wait (b); + if (res != 0 && res != PTHREAD_BARRIER_SERIAL_THREAD) + abort (); +} + +static void * +thread_worker (void *arg) +{ + barrier_wait (&barrier); + return NULL; +} + +void +breakpt (void) +{ + /* Nothing. */ +} + +int +main (void) +{ + void *handle; + void (*func)(int); + pthread_t thread; + + if (pthread_barrier_init (&barrier, NULL, 2) != 0) + abort (); + + if (pthread_create (&thread, NULL, thread_worker, NULL) != 0) + abort (); + + breakpt (); + + /* Allow the worker thread to complete. */ + barrier_wait (&barrier); + + if (pthread_join (thread, NULL) != 0) + abort (); + + breakpt (); + + /* Now load the shared library. */ + handle = dlopen (SHLIB_NAME, RTLD_LAZY); + if (handle == NULL) + abort (); + + /* Find the function symbol. */ + func = (void (*)(int)) dlsym (handle, "foo"); + + /* Call the library function. */ + func (1); + + /* Unload the shared library. */ + if (dlclose (handle) != 0) + abort (); + + breakpt (); + + return 0; +} + diff --git a/gdb/testsuite/gdb.threads/del-pending-thread-bp.exp b/gdb/testsuite/gdb.threads/del-pending-thread-bp.exp new file mode 100644 index 00000000000..14a91a40d76 --- /dev/null +++ b/gdb/testsuite/gdb.threads/del-pending-thread-bp.exp @@ -0,0 +1,98 @@ +# Copyright 2023 Free Software Foundation, Inc. + +# This program is free software; you can redistribute it and/or modify +# it under the terms of the GNU General Public License as published by +# the Free Software Foundation; either version 3 of the License, or +# (at your option) any later version. +# +# This program is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +# GNU General Public License for more details. +# +# You should have received a copy of the GNU General Public License +# along with this program. If not, see . + +# This test checks that pending thread-specific breakpoints are +# correctly deleted when the thread the breakpoint is for goes out of +# scope. +# +# We also check that we can't create a pending thread-specific +# breakpoint for a non-existent thread. + +require allow_shlib_tests + +standard_testfile + +set libname $testfile-lib +set srcfile_lib $srcdir/$subdir/$libname.c +set binfile_lib [standard_output_file $libname.so] + +if { [gdb_compile_shlib $srcfile_lib $binfile_lib {}] != "" } { + untested "failed to compile shared library 1" + return -1 +} + +set binfile_lib_target [gdb_download_shlib $binfile_lib] + +if { [prepare_for_testing "failed to prepare" $testfile $srcfile \ + [list debug \ + additional_flags=-DSHLIB_NAME=\"$binfile_lib_target\" \ + shlib_load pthreads]] } { + return -1 +} + +gdb_locate_shlib $binfile_lib + +if ![runto_main] { + return 0 +} + +# Run until we have two threads. +gdb_breakpoint "breakpt" +gdb_continue_to_breakpoint "first breakpt call" + +# Confirm that we have a thread '2'. +gdb_test "info threads" "\r\n\\s+2\\s+\[^\r\n\]+" + +# Create a pending, thread-specific, breakpoint on 'foo'. +gdb_breakpoint "foo thread 2" allow-pending +set bpnum [get_integer_valueof "\$bpnum" "*INVALID*" \ + "get breakpoint number"] + +# Check we can't create a pending thread-specific breakpoint for a +# non-existent thread. +gdb_test "with breakpoint pending on -- break foo thread 99" \ + "Unknown thread 99\\." + +# Continue to 'breakpt' again. Don't use gdb_continue_to_breakpoint +# as we are looking for the thread exited and breakpoint deleted +# messages. +gdb_test "continue" \ + [multi_line \ + "Continuing\\." \ + "\\\[Thread \[^\r\n\]+ exited\\\]" \ + "Thread-specific breakpoint $bpnum deleted - thread 2 no longer in the thread list\\." \ + "" \ + "Thread 1 \[^\r\n\]+, breakpt \\(\\) at \[^\r\n\]+" \ + "$decimal\\s+\[^\r\n\]+"] \ + "second breakpt call" + +# Confirm breakpoint has been deleted. +gdb_test "info breakpoints $bpnum" \ + "No breakpoint, watchpoint, tracepoint, or catchpoint matching '$bpnum'\\." + +# Continue again. This will pass through 'foo'. We should not stop +# in 'foo', the breakpoint has been deleted. We should next stop in +# breakpt again. +gdb_test "continue" \ + [multi_line \ + "Continuing\\." \ + "" \ + "Thread 1 \[^\r\n\]+ hit Breakpoint $decimal, breakpt \\(\\) at \[^\r\n\]+" \ + "$decimal\\s+\[^\r\n\]+"] \ + "third breakpt call" +gdb_test "bt 1" \ + [multi_line \ + "#0\\s+breakpt \\(\\) at \[^\r\n\]+" \ + "\\(More stack frames follow\\.\\.\\.\\)"]