public inbox for gdb-patches@sourceware.org
 help / color / mirror / Atom feed
From: Tom de Vries <tdevries@suse.de>
To: Luis Machado <luis.machado@arm.com>, gdb-patches@sourceware.org
Subject: Re: [pushed] [gdb/testsuite] Use maint ignore-probes in gdb.base/longjmp.exp
Date: Wed, 8 Feb 2023 15:48:36 +0100	[thread overview]
Message-ID: <23f68beb-39c5-33c8-2d2d-f076192d2648@suse.de> (raw)
In-Reply-To: <ac22420a-1549-bc0e-94f5-9f3e3f4af107@arm.com>

On 2/8/23 14:27, Luis Machado wrote:
> Hi Tom,
> 
> Is the entire test supposed to PASS? I'm seeing the following on my 
> aarch64/Ubuntu 22.04 setup:
> 
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 2: next over 
> call_longjmp (the program is no longer running)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 2: next over setjmp 
> (the program is no longer running)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 2: setup: breakpoint 
> at pattern start (got interactive prompt)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 2: setup: breakpoint 
> at safety net (got interactive prompt)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 2: setup: continue to 
> breakpoint at pattern start (the program exited)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 3: next over pattern 
> (the program is no longer running)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 3: setup: breakpoint 
> at pattern start (got interactive prompt)
> FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 3: setup: continue to 
> breakpoint at pattern start (the program is no longer running)
> 
> Maybe something is genuinely broken for aarch64 though, or I'm missing 
> some packages/debuginfo.

Hi,

I just ran this test-case on openSUSE Leap 15.4 aarch64, no problems found.

> 
> This part in particular seems a bit off:
> 
> (gdb) break 63
> No line 63 in the current file.
> Make breakpoint pending on future shared library load? (y or [n]) n
> (gdb) FAIL: gdb.base/longjmp.exp: with_probes=0: pattern 2: setup: 
> breakpoint at pattern start (got interactive prompt)
> 

I'd focus on the first FAIL.  Why is the safety net breakpoint not 
triggering?

Thanks,
- Tom

> On 2/8/23 12:46, Tom de Vries via Gdb-patches wrote:
>> Test-case gdb.base/longjmp.exp handles both the case that there is a libc
>> longjmp probe, and the case that there isn't.
>>
>> However, it only tests one of the two cases.
>>
>> Use maint ignore-probes to test both cases, if possible.
>>
>> Tested on x86_64-linux.
>> ---
>>   gdb/testsuite/gdb.base/longjmp.exp | 295 +++++++++++++++--------------
>>   1 file changed, 157 insertions(+), 138 deletions(-)
>>
>> diff --git a/gdb/testsuite/gdb.base/longjmp.exp 
>> b/gdb/testsuite/gdb.base/longjmp.exp
>> index 27a60bc4502..96d9c1c8059 100644
>> --- a/gdb/testsuite/gdb.base/longjmp.exp
>> +++ b/gdb/testsuite/gdb.base/longjmp.exp
>> @@ -25,181 +25,200 @@ if  { [gdb_compile 
>> "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {deb
>>        return -1
>>   }
>> -clean_restart ${binfile}
>> +proc do_test { with_probes } {
>> +    clean_restart ${::binfile}
>> -if {![runto_main]} {
>> -   return 0
>> -}
>> +    if { !$with_probes } {
>> +    gdb_test "maint ignore-probes libc ^longjmp$"
>> +    }
>> -# With a libc with probes, all tests should pass.
>> -#
>> -# Without probes, we can still set a break on longjmp, but getting 
>> the longjmp
>> -# target may not work, in the following cases:
>> -# - gdbarch_get_longjmp_target_p (gdbarch) == 0: not implemented.
>> -# - gdbarch_get_longjmp_target (gdbarch) == 0: for instance on amd64 if
>> -#   tdep->jb_pc_offset == -1.
>> -# - gdbarch_get_longjmp_target (gdbarch) != 0: if we have a glibc with
>> -#   pointer mangling ( 
>> https://sourceware.org/glibc/wiki/PointerEncryption )
>> -#   then we retrieve a mangled longjmp target that needs to be 
>> demangled.
>> -#   For instance on amd64 with target board unix/-m32.
>> -#
>> -# Pointer demangling is currently not implemented for any target.
>> -# For the amd64 case, this would require copying for instance this:
>> -#   48 c1 ca 11             ror    $0x11,%rdx
>> -#   64 48 33 14 25 30 00    xor    %fs:0x30,%rdx
>> -# into a scratch space, save the register set, set %rdx to the mangled
>> -# longjmp target, displaced-step through the two insn and read the
>> -# demangled longjmp target from %rdx, and restore the register set.
>> -#
>> -# The failure mode in the first two cases is that the next degrades 
>> into a
>> -# continue.  The failure mode in the latter case is a failure to set a
>> -# breakpoint (matched by re_cannot_insert_bp) and a stop in longjmp.
>> -#
>> -# We detect the different failure modes and kfail these.
>> +    if {![runto_main]} {
>> +    return 0
>> +    }
>> -set have_longjmp_probe 0
>> -gdb_test_multiple "info probes stap libc ^longjmp$" "" {
>> -    -re -wrap "No probes matched\\." {
>> -    pass $gdb_test_name
>> +    # With a libc with probes, all tests should pass.
>> +    #
>> +    # Without probes, we can still set a break on longjmp, but 
>> getting the longjmp
>> +    # target may not work, in the following cases:
>> +    # - gdbarch_get_longjmp_target_p (gdbarch) == 0: not implemented.
>> +    # - gdbarch_get_longjmp_target (gdbarch) == 0: for instance on 
>> amd64 if
>> +    #   tdep->jb_pc_offset == -1.
>> +    # - gdbarch_get_longjmp_target (gdbarch) != 0: if we have a glibc 
>> with
>> +    #   pointer mangling ( 
>> https://sourceware.org/glibc/wiki/PointerEncryption )
>> +    #   then we retrieve a mangled longjmp target that needs to be 
>> demangled.
>> +    #   For instance on amd64 with target board unix/-m32.
>> +    #
>> +    # Pointer demangling is currently not implemented for any target.
>> +    # For the amd64 case, this would require copying for instance this:
>> +    #   48 c1 ca 11             ror    $0x11,%rdx
>> +    #   64 48 33 14 25 30 00    xor    %fs:0x30,%rdx
>> +    # into a scratch space, save the register set, set %rdx to the 
>> mangled
>> +    # longjmp target, displaced-step through the two insn and read the
>> +    # demangled longjmp target from %rdx, and restore the register set.
>> +    #
>> +    # The failure mode in the first two cases is that the next 
>> degrades into a
>> +    # continue.  The failure mode in the latter case is a failure to 
>> set a
>> +    # breakpoint (matched by re_cannot_insert_bp) and a stop in longjmp.
>> +    #
>> +    # We detect the different failure modes and kfail these.
>> +
>> +    set have_longjmp_probe 0
>> +    gdb_test_multiple "info probes stap libc ^longjmp$" "" {
>> +    -re -wrap "No probes matched\\." {
>> +        pass $gdb_test_name
>> +    }
>> +    -re -wrap "\r\nstap\[ \t\]+libc\[ \t\]+longjmp\[ \t\]+.*" {
>> +        pass $gdb_test_name
>> +        set have_longjmp_probe 1
>> +    }
>>       }
>> -    -re -wrap "\r\nstap\[ \t\]+libc\[ \t\]+longjmp\[ \t\]+.*" {
>> -    pass $gdb_test_name
>> -    set have_longjmp_probe 1
>> +
>> +    if { $with_probes } {
>> +    if { !$have_longjmp_probe } {
>> +        unsupported "longjmp probe required"
>> +        return
>> +    }
>> +    } else {
>> +    gdb_assert { !$have_longjmp_probe }
>>       }
>> -}
>> -set bp_miss_step_1 [gdb_get_line_number "miss_step_1"]
>> -set bp_miss_step_2 [gdb_get_line_number "miss_step_2"]
>> +    set bp_miss_step_1 [gdb_get_line_number "miss_step_1"]
>> +    set bp_miss_step_2 [gdb_get_line_number "miss_step_2"]
>> -set bp_start_test_1 [gdb_get_line_number "patt1"]
>> -set bp_start_test_2 [gdb_get_line_number "patt2"]
>> -set bp_start_test_3 [gdb_get_line_number "patt3"]
>> +    set bp_start_test_1 [gdb_get_line_number "patt1"]
>> +    set bp_start_test_2 [gdb_get_line_number "patt2"]
>> +    set bp_start_test_3 [gdb_get_line_number "patt3"]
>> -set re_cannot_insert_bp \
>> -    [multi_line \
>> -     "Warning:" \
>> -     "Cannot insert breakpoint $decimal\\." \
>> -     "Cannot access memory at address $hex"]
>> +    set re_cannot_insert_bp \
>> +    [multi_line \
>> +         "Warning:" \
>> +         "Cannot insert breakpoint $::decimal\\." \
>> +         "Cannot access memory at address $::hex"]
>> -#
>> -# Pattern 1 - simple longjmp.
>> -#
>> +    #
>> +    # Pattern 1 - simple longjmp.
>> +    #
>> -with_test_prefix "pattern 1" {
>> +    with_test_prefix "pattern 1" {
>> -    with_test_prefix setup {
>> -    delete_breakpoints
>> +    with_test_prefix setup {
>> +        delete_breakpoints
>> -    gdb_test "break $bp_start_test_1" \
>> -        "Breakpoint.*at.* file .*$srcfile, line.*$bp_start_test_1.*" \
>> -        "breakpoint at pattern start"
>> -    gdb_test "continue" "patt1.*" "continue to breakpoint at pattern 
>> start"
>> +        gdb_test "break $bp_start_test_1" \
>> +        "Breakpoint.*at.* file .*$::srcfile, line.*$bp_start_test_1.*" \
>> +        "breakpoint at pattern start"
>> +        gdb_test "continue" "patt1.*" "continue to breakpoint at 
>> pattern start"
>> -    # set safe-net break
>> -    gdb_test "break $bp_miss_step_1" \
>> -        "Breakpoint.*at.* file .*$srcfile, line.*$bp_miss_step_1.*" \
>> -        "breakpoint at safety net"
>> -    }
>> +        # set safe-net break
>> +        gdb_test "break $bp_miss_step_1" \
>> +        "Breakpoint.*at.* file .*$::srcfile, line.*$bp_miss_step_1.*" \
>> +        "breakpoint at safety net"
>> +    }
>> -    gdb_test "next" "longjmps\\+\\+;.*" "next over setjmp"
>> -    gdb_test "next" "longjmp \\(env, 1\\);.*" "next to longjmp"
>> +    gdb_test "next" "longjmps\\+\\+;.*" "next over setjmp"
>> +    gdb_test "next" "longjmp \\(env, 1\\);.*" "next to longjmp"
>> -    set msg "next over longjmp"
>> -    gdb_test_multiple "next" $msg {
>> -    -re ".*patt1.*$gdb_prompt $" {
>> -        pass $msg
>> -        gdb_test "next" "resumes\\+\\+.*" "next into else block"
>> -        gdb_test "next" "miss_step_1.*" "next into safety net"
>> -    }
>> -    -re "miss_step_1.*$gdb_prompt $" {
>> -        if { $have_longjmp_probe } {
>> -        fail $gdb_test_name
>> -        } else {
>> -        kfail $gdb_test_name "gdb/26967"
>> +    set msg "next over longjmp"
>> +    gdb_test_multiple "next" $msg {
>> +        -re ".*patt1.*$::gdb_prompt $" {
>> +        pass $msg
>> +        gdb_test "next" "resumes\\+\\+.*" "next into else block"
>> +        gdb_test "next" "miss_step_1.*" "next into safety net"
>>           }
>> -    }
>> -    -re -wrap "\r\n$re_cannot_insert_bp\r\n.*" {
>> -        if { $have_longjmp_probe } {
>> -        fail $gdb_test_name
>> -        } else {
>> -        kfail $gdb_test_name "gdb/26967"
>> +        -re "miss_step_1.*$::gdb_prompt $" {
>> +        if { $have_longjmp_probe } {
>> +            fail $gdb_test_name
>> +        } else {
>> +            kfail $gdb_test_name "gdb/26967"
>> +        }
>> +        }
>> +        -re -wrap "\r\n$re_cannot_insert_bp\r\n.*" {
>> +        if { $have_longjmp_probe } {
>> +            fail $gdb_test_name
>> +        } else {
>> +            kfail $gdb_test_name "gdb/26967"
>> +        }
>>           }
>>       }
>>       }
>> -}
>> -#
>> -# Pattern 2 - longjmp from an inner function.
>> -#
>> +    #
>> +    # Pattern 2 - longjmp from an inner function.
>> +    #
>> -with_test_prefix "pattern 2" {
>> +    with_test_prefix "pattern 2" {
>> -    with_test_prefix setup {
>> -    delete_breakpoints
>> +    with_test_prefix setup {
>> +        delete_breakpoints
>> -    gdb_test "break $bp_start_test_2" \
>> -        "Breakpoint.*at.* file .*$srcfile, line.*$bp_start_test_2.*" \
>> -        "breakpoint at pattern start"
>> -    gdb_test "continue" "patt2.*" "continue to breakpoint at pattern 
>> start"
>> +        gdb_test "break $bp_start_test_2" \
>> +        "Breakpoint.*at.* file .*$::srcfile, line.*$bp_start_test_2.*" \
>> +        "breakpoint at pattern start"
>> +        gdb_test "continue" "patt2.*" "continue to breakpoint at 
>> pattern start"
>> -    # set safe-net break
>> -    gdb_test "break $bp_miss_step_2" \
>> -        "Breakpoint.*at.* file .*$srcfile, line.*$bp_miss_step_2.*" \
>> -        "breakpoint at safety net"
>> -    }
>> +        # set safe-net break
>> +        gdb_test "break $bp_miss_step_2" \
>> +        "Breakpoint.*at.* file .*$::srcfile, line.*$bp_miss_step_2.*" \
>> +        "breakpoint at safety net"
>> +    }
>> -    gdb_test "next" "call_longjmp.*" "next over setjmp"
>> +    gdb_test "next" "call_longjmp.*" "next over setjmp"
>> -    set msg "next over call_longjmp"
>> -    gdb_test_multiple "next" $msg {
>> -    -re ".*patt2.*$gdb_prompt $" {
>> -        pass $msg
>> +    set msg "next over call_longjmp"
>> +    gdb_test_multiple "next" $msg {
>> +        -re ".*patt2.*$::gdb_prompt $" {
>> +        pass $msg
>> -        gdb_test "next" "resumes\\+\\+.*" "next into else block"
>> -        gdb_test "next" "miss_step_2.*" "next into safety net"
>> -    }
>> -    -re "miss_step_2.*$gdb_prompt $" {
>> -        if { $have_longjmp_probe } {
>> -        fail $gdb_test_name
>> -        } else {
>> -        kfail $gdb_test_name "gdb/26967"
>> +        gdb_test "next" "resumes\\+\\+.*" "next into else block"
>> +        gdb_test "next" "miss_step_2.*" "next into safety net"
>>           }
>> -    }
>> -    -re -wrap "\r\n$re_cannot_insert_bp\r\n.*" {
>> -        if { $have_longjmp_probe } {
>> -        fail $gdb_test_name
>> -        } else {
>> -        kfail $gdb_test_name "gdb/26967"
>> +        -re "miss_step_2.*$::gdb_prompt $" {
>> +        if { $have_longjmp_probe } {
>> +            fail $gdb_test_name
>> +        } else {
>> +            kfail $gdb_test_name "gdb/26967"
>> +        }
>> +        }
>> +        -re -wrap "\r\n$re_cannot_insert_bp\r\n.*" {
>> +        if { $have_longjmp_probe } {
>> +            fail $gdb_test_name
>> +        } else {
>> +            kfail $gdb_test_name "gdb/26967"
>> +        }
>>           }
>>       }
>>       }
>> -}
>> -#
>> -# Pattern 3 - setjmp/longjmp inside stepped-over function.
>> -#
>> +    #
>> +    # Pattern 3 - setjmp/longjmp inside stepped-over function.
>> +    #
>> -with_test_prefix "pattern 3" {
>> +    with_test_prefix "pattern 3" {
>> -    with_test_prefix setup {
>> -    delete_breakpoints
>> +    with_test_prefix setup {
>> +        delete_breakpoints
>> -    gdb_test "break $bp_start_test_3" \
>> -        "Breakpoint.*at.* file .*$srcfile, line.*$bp_start_test_3.*" \
>> -        "breakpoint at pattern start"
>> -    gdb_test "continue" "patt3.*" "continue to breakpoint at pattern 
>> start"
>> -    }
>> -
>> -    gdb_test_multiple "next" "next over pattern" {
>> -    -re -wrap "longjmp caught.*" {
>> -        pass $gdb_test_name
>> +        gdb_test "break $bp_start_test_3" \
>> +        "Breakpoint.*at.* file .*$::srcfile, line.*$bp_start_test_3.*" \
>> +        "breakpoint at pattern start"
>> +        gdb_test "continue" "patt3.*" "continue to breakpoint at 
>> pattern start"
>>       }
>> -    -re -wrap "\r\n$re_cannot_insert_bp\r\n.*" {
>> -        if { $have_longjmp_probe } {
>> -        fail $gdb_test_name
>> -        } else {
>> -        kfail $gdb_test_name "gdb/26967"
>> +
>> +    gdb_test_multiple "next" "next over pattern" {
>> +        -re -wrap "longjmp caught.*" {
>> +        pass $gdb_test_name
>> +        }
>> +        -re -wrap "\r\n$re_cannot_insert_bp\r\n.*" {
>> +        if { $have_longjmp_probe } {
>> +            fail $gdb_test_name
>> +        } else {
>> +            kfail $gdb_test_name "gdb/26967"
>> +        }
>>           }
>>       }
>>       }
>>   }
>> +
>> +foreach_with_prefix with_probes { 0 1 } {
>> +    do_test $with_probes
>> +}
>>
>> base-commit: 0ab9328277ce12014ce05ff139eba7a6230516fd
> 


  reply	other threads:[~2023-02-08 14:48 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-02-08 12:46 Tom de Vries
2023-02-08 13:27 ` Luis Machado
2023-02-08 14:48   ` Tom de Vries [this message]
2023-02-08 14:51     ` Luis Machado
2023-02-08 15:38       ` Tom de Vries
2023-02-08 18:06         ` Luis Machado
2023-02-08 20:36           ` Tom de Vries
2023-02-09 10:37             ` Luis Machado
2023-02-09 11:58               ` Luis Machado
2023-02-09 12:19                 ` Tom de Vries
2023-02-09 14:34                   ` Luis Machado
2023-02-09 16:44                     ` Luis Machado
2023-02-10 11:09                       ` Tom de Vries

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=23f68beb-39c5-33c8-2d2d-f076192d2648@suse.de \
    --to=tdevries@suse.de \
    --cc=gdb-patches@sourceware.org \
    --cc=luis.machado@arm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).