valgrind-developers — Discussion list for developers of valgrind

[Valgrind-developers] [COMMITTED] Add valgrind [try] builders for riscv (starfive and bpi-f3)

[Mark W. <ma...@kl...>]

---
 builder/master.cfg | 36 ++++++++++++++++++++++++++++++++++++
 1 file changed, 36 insertions(+)

diff --git a/builder/master.cfg b/builder/master.cfg
index 7a55e284301a..8fec93d3b09c 100644
--- a/builder/master.cfg
+++ b/builder/master.cfg
@@ -874,6 +874,8 @@ valgrind_scheduler = schedulers.SingleBranchScheduler(
                       "valgrind-ibm-power9",
 #                      "valgrind-ibm-power10",
                       "valgrind-fedora-ppc64le",
+                      "valgrind-fedora-riscv",
+                      "valgrind-ubuntu-riscv",
                       "valgrind-fedora-s390x",
                       "valgrind-debian-ppc64",
 #                      "valgrind-debian-arm64",
@@ -892,6 +894,8 @@ valgrind_try_scheduler = schedulers.AnyBranchScheduler(
 #                      "valgrind-try-ibm-power10",
                       "valgrind-try-fedora-x86_64",
                       "valgrind-try-fedora-ppc64le",
+                      "valgrind-try-fedora-riscv",
+                      "valgrind-try-ubuntu-riscv",
                       "valgrind-try-fedora-s390x",
                       "valgrind-try-debian-ppc64",
 #                      "valgrind-try-debian-arm64",
@@ -5139,6 +5143,38 @@ valgrind_try_fedora_ppc64le_builder = util.BuilderConfig(
         factory=valgrind_make_check_factory)
 c['builders'].append(valgrind_try_fedora_ppc64le_builder)
 
+valgrind_ubuntu_riscv_builder = util.BuilderConfig(
+        name="valgrind-ubuntu-riscv",
+        workernames=starfive_workers,
+        tags=["valgrind", "ubuntu", "riscv"],
+        collapseRequests=True,
+        factory=valgrind_make_check_factory)
+c['builders'].append(valgrind_ubuntu_riscv_builder)
+
+valgrind_try_ubuntu_riscv_builder = util.BuilderConfig(
+        name="valgrind-try-ubuntu-riscv",
+        workernames=starfive_workers,
+        tags=["valgrind-try", "ubuntu", "riscv"],
+        collapseRequests=True,
+        factory=valgrind_make_check_factory)
+c['builders'].append(valgrind_try_ubuntu_riscv_builder)
+
+valgrind_fedora_riscv_builder = util.BuilderConfig(
+        name="valgrind-fedora-riscv",
+        workernames=bpi_f3_workers,
+        tags=["valgrind", "fedora", "riscv"],
+        collapseRequests=True,
+        factory=valgrind_make_check_factory)
+c['builders'].append(valgrind_fedora_riscv_builder)
+
+valgrind_try_fedora_riscv_builder = util.BuilderConfig(
+        name="valgrind-try-fedora-riscv",
+        workernames=bpi_f3_workers,
+        tags=["valgrind-try", "fedora", "riscv"],
+        collapseRequests=True,
+        factory=valgrind_make_check_factory)
+c['builders'].append(valgrind_try_fedora_riscv_builder)
+
 valgrind_ibm_power9_builder = util.BuilderConfig(
         name="valgrind-ibm-power9",
         workernames=["ibm_power9"],
-- 
2.48.1

[Valgrind-developers] [valgrind] Illumos regtest: update bug487993 testcase

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=350f0028fcd8ae6db5b0500ed32fc9aa75ab9e7e

commit 350f0028fcd8ae6db5b0500ed32fc9aa75ab9e7e
Author: Paul Floyd <pj...@wa...>
Date:   Sun Feb 16 09:36:54 2025 +0100

    Illumos regtest: update bug487993 testcase
    
    As I said in bugzilla, "Not tested on Solaris". Now I've tested it
    and Illumos libc doesn't use 16byte alignment (or at least the assert
    that alignment is 16bytes fails).

Diff:
---
 memcheck/tests/x86/bug487993.cpp    | 7 ++++++-
 memcheck/tests/x86/bug487993.vgtest | 2 +-
 2 files changed, 7 insertions(+), 2 deletions(-)

diff --git a/memcheck/tests/x86/bug487993.cpp b/memcheck/tests/x86/bug487993.cpp
index 27f43d306c..e1678c9538 100644
--- a/memcheck/tests/x86/bug487993.cpp
+++ b/memcheck/tests/x86/bug487993.cpp
@@ -12,9 +12,14 @@ struct Test
 
 int main()
 {
+#if defined(__sun__)
+    const auto alignment_mask{0x7UL};
+#else
+    const auto alignment_mask{0xfUL};
+#endif
     std::unique_ptr<Test> test = std::make_unique<Test>();
     //std::cerr << "test " << test.get() << ' ' << (reinterpret_cast<size_t>(test.get()) & 0xfUL) << '\n';
-    assert((reinterpret_cast<size_t>(test.get()) & 0xfUL) == 0);
+    assert((reinterpret_cast<size_t>(test.get()) & alignment_mask) == 0);
 }
 
     
diff --git a/memcheck/tests/x86/bug487993.vgtest b/memcheck/tests/x86/bug487993.vgtest
index a5f00a4821..ca5db1515d 100644
--- a/memcheck/tests/x86/bug487993.vgtest
+++ b/memcheck/tests/x86/bug487993.vgtest
@@ -1,3 +1,3 @@
-prereq: test -e bug487993 && ! ../../../tests/os_test solaris
+prereq: test -e bug487993
 prog: bug487993
 vgopts: -q

[Valgrind-developers] [valgrind] Illumos regtest: update memcheck dw4 expeected

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=e9645efebc16724c96c83c8b75f641fc32ad993f

commit e9645efebc16724c96c83c8b75f641fc32ad993f
Author: Paul Floyd <pj...@wa...>
Date:   Sun Feb 16 08:43:19 2025 +0100

    Illumos regtest: update memcheck dw4 expeected
    
    The exp-solaris version was the same as the exp one.

Diff:
---
 memcheck/tests/dw4.stderr.exp | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/memcheck/tests/dw4.stderr.exp b/memcheck/tests/dw4.stderr.exp
index 917dcb57e3..e4b523f982 100644
--- a/memcheck/tests/dw4.stderr.exp
+++ b/memcheck/tests/dw4.stderr.exp
@@ -35,12 +35,12 @@ Unaddressable byte(s) found during client check request
 Uninitialised byte(s) found during client check request
    at 0x........: croak (dw4.c:32)
    by 0x........: main (dw4.c:89)
- Address 0x........ is in the brk data segment 0x........-0x........
+ Address 0x........ is in a rw- mapped file /export/home/paulf/valgrind/memcheck/tests/dw4 segment
 
 Uninitialised byte(s) found during client check request
    at 0x........: croak (dw4.c:32)
    by 0x........: main (dw4.c:91)
- Address 0x........ is in the brk data segment 0x........-0x........
+ Address 0x........ is in a rw- mapped file /export/home/paulf/valgrind/memcheck/tests/dw4 segment
 
 Unaddressable byte(s) found during client check request
    at 0x........: croak (dw4.c:25)

[Valgrind-developers] [valgrind] Illumos regtest: add a suppression for leak_cpp_intertior libstdc++.supp

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=4a2944627ef6087291c67ec07b45c9c3d7c35e64

commit 4a2944627ef6087291c67ec07b45c9c3d7c35e64
Author: Paul Floyd <pj...@wa...>
Date:   Sun Feb 16 08:38:58 2025 +0100

    Illumos regtest: add a suppression for leak_cpp_intertior libstdc++.supp

Diff:
---
 memcheck/tests/libstdc++.supp | 12 ++++++++++++
 1 file changed, 12 insertions(+)

diff --git a/memcheck/tests/libstdc++.supp b/memcheck/tests/libstdc++.supp
index 525b62832f..bcfb4a0648 100644
--- a/memcheck/tests/libstdc++.supp
+++ b/memcheck/tests/libstdc++.supp
@@ -122,3 +122,15 @@
    fun:_setup
    fun:_rt_boot
 }
+{
+   Illumos-leaks-g++-malloc
+   Memcheck:Leak
+   match-leak-kinds: reachable
+   fun:malloc
+   fun:_GLOBAL__sub_I_eh_alloc.cc
+   fun:call_array
+   fun:call_init
+   fun:setup
+   fun:_setup
+   fun:_rt_boot
+}

[Valgrind-developers] [valgrind] Illumos regtest: getrandom uses a different syscall number from Solaris

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=a1f0a7daeb3ef4c98ad54821419fe02597592265

commit a1f0a7daeb3ef4c98ad54821419fe02597592265
Author: Paul Floyd <pj...@wa...>
Date:   Sun Feb 16 07:57:43 2025 +0100

    Illumos regtest: getrandom uses a different syscall number from Solaris
    
    It doesn't look like the syscall interface changed, just the number.

Diff:
---
 coregrind/m_syswrap/syswrap-solaris.c                    |  4 ++--
 memcheck/tests/solaris/Makefile.am                       |  1 +
 .../tests/solaris/scalar_getrandom.stderr.exp-illumos    | 16 ++++++++++++++++
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/coregrind/m_syswrap/syswrap-solaris.c b/coregrind/m_syswrap/syswrap-solaris.c
index faeea15f54..3da0d216cd 100644
--- a/coregrind/m_syswrap/syswrap-solaris.c
+++ b/coregrind/m_syswrap/syswrap-solaris.c
@@ -10947,7 +10947,7 @@ static SyscallTableEntry syscall_table[] = {
    SOLXY(__NR_uuidsys,              sys_uuidsys),               /* 124 */
 #endif /* SOLARIS_UUIDSYS_SYSCALL */
 #if defined(HAVE_MREMAP)
-   GENX_(__NR_mremap,               sys_mremap),                /* 126 */
+   GENX_(__NR_mremap,               sys_mremap),                /* 126 (Solaris only, not Illumos) */
 #endif /* HAVE_MREMAP */
    SOLX_(__NR_mmapobj,              sys_mmapobj),               /* 127 */
    GENX_(__NR_setrlimit,            sys_setrlimit),             /* 128 */
@@ -10968,7 +10968,7 @@ static SyscallTableEntry syscall_table[] = {
    SOLX_(__NR_seteuid,              sys_seteuid),               /* 141 */
    SOLX_(__NR_forksys,              sys_forksys),               /* 142 */
 #if defined(SOLARIS_GETRANDOM_SYSCALL)
-   SOLXY(__NR_getrandom,            sys_getrandom),             /* 143 */
+   SOLXY(__NR_getrandom,            sys_getrandom),             /* 143 (Solaris) 126 (Illumos) */
 #endif /* SOLARIS_GETRANDOM_SYSCALL */
    SOLXY(__NR_sigtimedwait,         sys_sigtimedwait),          /* 144 */
    SOLX_(__NR_yield,                sys_yield),                 /* 146 */
diff --git a/memcheck/tests/solaris/Makefile.am b/memcheck/tests/solaris/Makefile.am
index 7cf1ac07ce..0c0d551ed0 100644
--- a/memcheck/tests/solaris/Makefile.am
+++ b/memcheck/tests/solaris/Makefile.am
@@ -34,6 +34,7 @@ EXTRA_DIST = \
 	scalar_auditon_stat.stderr.exp scalar_auditon_stat.stdout.exp scalar_auditon_stat.vgtest \
 	scalar_frealpathat.stderr.exp scalar_frealpathat.stdout.exp scalar_frealpathat.vgtest \
 	scalar_getrandom.stderr.exp scalar_getrandom.stdout.exp scalar_getrandom.vgtest \
+	scalar_getrandom.stderr.exp-illumos \
 	scalar_ioctl.stderr.exp scalar_ioctl.stdout.exp scalar_ioctl.vgtest \
 	scalar_lwp_kill.stderr.exp scalar_lwp_kill.stdout.exp scalar_lwp_kill.vgtest \
 	scalar_lwp_name.stderr.exp scalar_lwp_name.stdout.exp scalar_lwp_name.vgtest \
diff --git a/memcheck/tests/solaris/scalar_getrandom.stderr.exp-illumos b/memcheck/tests/solaris/scalar_getrandom.stderr.exp-illumos
new file mode 100644
index 0000000000..7e84a4849f
--- /dev/null
+++ b/memcheck/tests/solaris/scalar_getrandom.stderr.exp-illumos
@@ -0,0 +1,16 @@
+---------------------------------------------------------
+126:           SYS_getrandom (getrandom) 3s 1m
+---------------------------------------------------------
+Syscall param getrandom(buf) contains uninitialised byte(s)
+   ...
+
+Syscall param getrandom(buflen) contains uninitialised byte(s)
+   ...
+
+Syscall param getrandom(flags) contains uninitialised byte(s)
+   ...
+
+Syscall param getrandom(buf) points to unaddressable byte(s)
+   ...
+ Address 0x........ is not stack'd, malloc'd or (recently) free'd
+

[Valgrind-developers] [valgrind] Illumos regtest: improve filters

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=8dc7dcf1fa595e5c21b34f73e45190e6d2b5f09d

commit 8dc7dcf1fa595e5c21b34f73e45190e6d2b5f09d
Author: Paul Floyd <pj...@wa...>
Date:   Sat Feb 15 20:30:11 2025 +0100

    Illumos regtest: improve filters
    
    filter_fdleak - move the Solaris filters earlier, there is a multiline awk filter
    that was being affected by the arm/FreeBSD filters
    
    coredump filter is different between Solaris and Illumos

Diff:
---
 none/tests/filter_fdleak                             | 19 ++++++++++---------
 none/tests/solaris/filter_coredump_many_threads_post |  3 ++-
 2 files changed, 12 insertions(+), 10 deletions(-)

diff --git a/none/tests/filter_fdleak b/none/tests/filter_fdleak
index 260c6e24c0..f84008b6db 100755
--- a/none/tests/filter_fdleak
+++ b/none/tests/filter_fdleak
@@ -16,6 +16,16 @@ perl -p -e 's/^Open file descriptor [0-9]*:$/Open file descriptor ...:/' |
 perl -p -e 's/File descriptor [0-9]*: .* is already closed/File descriptor ...: ... is already closed/' |
 perl -p -e 's/127.0.0.1:[0-9]*/127.0.0.1:.../g' |
 perl -p -e 's/socket\.c:[1-9][0-9]*/in \/...libc.../' |
+
+# Solaris specific fdleak filters
+perl -p -e 's/ __close / close /' |
+perl -p -e 's/ __write / write /' |
+sed '/ _so_socket /d;' |
+awk '/ socket /{sub(/ by /, " at "); print; next}{print}' |
+awk '/ at .* _syscall6 /{getline; getline; sub(/ by /, " at "); print; next}{print}' |
+sed '/__open/d' |
+sed 's/__systemcall6/creat/' |
+
 # arm systems substitute open for creat
 perl -p -e 's/open \(open64\.c:[1-9][0-9]*\)/creat (in \/...libc...)/' |
 perl -p -e "s/: open \(/: creat (/" |
@@ -28,15 +38,6 @@ sed '/by 0x........: close (in \/...libc...)/d' |
 sed '/by 0x........: creat (in \/...libc...)/d' |
 sed '/by 0x........: write (in \/...libc...)/d' |
 
-# Solaris specific fdleak filters
-perl -p -e 's/ __close / close /' |
-perl -p -e 's/ __write / write /' |
-sed '/ _so_socket /d;' |
-awk '/ socket /{sub(/ by /, " at "); print; next}{print}' |
-awk '/ at .* _syscall6 /{getline; getline; sub(/ by /, " at "); print; next}{print}' |
-sed '/__open/d' |
-sed 's/__systemcall6/creat/' |
-
 # Stack traces showing where fds were created have lots of variations:
 # different numbers of entries, line numbers or not depending on whether
 # libc debug info is present, 'main' vs. '(below main)', etc.  
diff --git a/none/tests/solaris/filter_coredump_many_threads_post b/none/tests/solaris/filter_coredump_many_threads_post
index 2d033700ef..e8284d1977 100755
--- a/none/tests/solaris/filter_coredump_many_threads_post
+++ b/none/tests/solaris/filter_coredump_many_threads_post
@@ -16,4 +16,5 @@ perl -p -e 's/^(THREAD)\s+(STATE)\s+(SOBJ)\s+(COUNT)\s*$/$1 $2 $3 $4\n/' | \
 perl -p -e 's/^(\d+)\s+(UNPARKED)\s+(\S+)\s+(\d+)/$1 $2 $3 $4/g' | \
 perl -p -e 's/^\s*libc.*.so.1/libc.so.1/g' | \
 perl -p -e 's/\+0x[0-9A-Fa-f]+//g' |
-sed 's/^ *//g'
+sed 's/^ *//g' |
+sed '/_start_crt/d'

[Valgrind-developers] [valgrind] Illumos regtest: add a filter for crt start

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=01a6ea5b031e48f35748a3df4e37bdaa25d61264

commit 01a6ea5b031e48f35748a3df4e37bdaa25d61264
Author: Paul Floyd <pj...@wa...>
Date:   Sat Feb 15 16:25:23 2025 +0100

    Illumos regtest: add a filter for crt start
    
    Was causing a failure in one Massif test

Diff:
---
 tests/filter_libc | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/tests/filter_libc b/tests/filter_libc
index 160f1a4b4d..d3c875d90f 100755
--- a/tests/filter_libc
+++ b/tests/filter_libc
@@ -26,6 +26,9 @@ while (<>)
     s#_start1 \(src/lib/csu/i386/crt1_c.c.*#(below main)#;
     s#__libc_start1 \(in /...libc...\)#(below main)#;
 
+    # on Illumos
+    s#_start_crt \(in .*#(below main)#;
+
     # filter out the exact libc-start.c:### line number. (ppc64*)
     s/\(libc-start.c:[0-9]*\)$/(in \/...libc...)/;

[Valgrind-developers] [valgrind] Bug 496370 - Illumos: signal handling is broken

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=ba5e30347fa5691cc9b0c4c620ee41daa135e610

commit ba5e30347fa5691cc9b0c4c620ee41daa135e610
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 14 20:45:02 2025 +0100

    Bug 496370 - Illumos: signal handling is broken
    
    This isn't a great fix, it just turns off optimization for
    a couple of signal frame functions. If ever I have time I'll try
    to find out out which part of -O1 is responsible, and maybe from
    that also exactly what part of the code.

Diff:
---
 NEWS                                  | 2 ++
 coregrind/m_syswrap/syswrap-solaris.c | 5 +++++
 2 files changed, 7 insertions(+)

diff --git a/NEWS b/NEWS
index 1c627c7744..fb7642bb19 100644
--- a/NEWS
+++ b/NEWS
@@ -40,6 +40,7 @@ are not entered into bugzilla tend to get forgotten about or ignored.
 494327  Crash when running Helgrind built with #define TRACE_PTH_FNS 1
 494337  All threaded applications cause still holding lock errors
 495488  Add FreeBSD getrlimitusage syscall wrapper
+496370  Illumos: signal handling is broken
 496571  False positive for null key passed to bpf_map_get_next_key syscall.
 469782  Valgrind does not support zstd-compressed debug sections
 497130  Recognize new DWARF5 DW_LANG constants
@@ -53,6 +54,7 @@ are not entered into bugzilla tend to get forgotten about or ignored.
 499183  FreeBSD: differences in avx-vmovq output
 499212  mmap() with MAP_ALIGNED() returns unaligned pointer
 
+
 To see details of a given bug, visit
   https://bugs.kde.org/show_bug.cgi?id=XXXXXX
 where XXXXXX is the bug number as listed above.
diff --git a/coregrind/m_syswrap/syswrap-solaris.c b/coregrind/m_syswrap/syswrap-solaris.c
index 9abeb85001..faeea15f54 100644
--- a/coregrind/m_syswrap/syswrap-solaris.c
+++ b/coregrind/m_syswrap/syswrap-solaris.c
@@ -511,6 +511,9 @@ ULong ML_(fletcher64)(UInt *buf, SizeT blocks)
    return (sum2 << 32) | sum1;
 }
 
+#pragma GCC push_options
+#pragma GCC optimize ("O0")
+
 /* Save a complete context (VCPU state, sigmask) of a given client thread
    into the vki_ucontext_t structure.  This structure is supposed to be
    allocated in the client memory, a caller must make sure that the memory can
@@ -666,6 +669,8 @@ void VG_(restore_context)(ThreadId tid, vki_ucontext_t *uc, CorePart part,
    }
 }
 
+#pragma GCC pop_options
+
 /* Set a client stack associated with a given thread id according to values
    passed in the vki_stack_t structure. */
 static void set_stack(ThreadId tid, vki_stack_t *st)

[Valgrind-developers] [valgrind] Soalris suppressions for pthread_once and gomp init

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=97e1ae53067200f6046a0e23f03392ecc9ed30e4

commit 97e1ae53067200f6046a0e23f03392ecc9ed30e4
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 14 20:42:33 2025 +0100

    Soalris suppressions for pthread_once and gomp init

Diff:
---
 solaris11.supp | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/solaris11.supp b/solaris11.supp
index ef8c5b5a42..e3ef0b7843 100644
--- a/solaris11.supp
+++ b/solaris11.supp
@@ -49,4 +49,17 @@
    fun:_ZNSt5mutex4lockEv
    fun:_ZNSt10lock_guardISt5mutexEC1ERS0_
 }
+{
+   Solaris:pthread_once
+   Helgrind:Race
+   fun:pthread_once
+}
+{
+   Solaris:gomp_init_task
+   drd:ConflictingAccess
+   fun:priority_queue_init
+   fun:gomp_init_task
+   fun:gomp_team_start
+   fun:GOMP_parallel
+}

[Valgrind-developers] [valgrind] Solaris regtest: add a suppression to helgrind/tests/bug392331.supp

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=612a84d112b70feea6b76d5a8f9a8c65c0cc52bb

commit 612a84d112b70feea6b76d5a8f9a8c65c0cc52bb
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 14 20:36:52 2025 +0100

    Solaris regtest: add a suppression to helgrind/tests/bug392331.supp

Diff:
---
 helgrind/tests/bug392331.supp | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/helgrind/tests/bug392331.supp b/helgrind/tests/bug392331.supp
index d76ec469e0..27081c771d 100644
--- a/helgrind/tests/bug392331.supp
+++ b/helgrind/tests/bug392331.supp
@@ -11,3 +11,9 @@
    fun:pthread_cond_signal_WRK
    fun:_ZNSt3__118condition_variable10notify_oneEv
 }
+{
+   Variation for Solaris
+   Helgrind:Dubious
+   fun:pthread_cond_signal_WRK
+   fun:cond_signal
+}

[Valgrind-developers] [valgrind] Solaris regtest: add a suppression for GCC 12 libstdc++ memcheck leak_cpp_interior

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=8e76b87ddc90e6fa7262c3ee08eb14405b7eaa18

commit 8e76b87ddc90e6fa7262c3ee08eb14405b7eaa18
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 14 19:34:27 2025 +0100

    Solaris regtest: add a suppression for GCC 12 libstdc++ memcheck leak_cpp_interior

Diff:
---
 memcheck/tests/libstdc++.supp | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/memcheck/tests/libstdc++.supp b/memcheck/tests/libstdc++.supp
index 2cac057305..525b62832f 100644
--- a/memcheck/tests/libstdc++.supp
+++ b/memcheck/tests/libstdc++.supp
@@ -109,3 +109,16 @@
    obj:/libexec/ld-elf*.so.1
    obj:/libexec/ld-elf*.so.1
 }
+{
+   Solaris-leaks-g++-malloc
+   Memcheck:Leak
+   match-leak-kinds: reachable
+   fun:malloc
+   fun:_GLOBAL__sub_I_eh_alloc.cc
+   obj:*/libstdc++.so*
+   fun:_init
+   fun:call_init
+   fun:setup
+   fun:_setup
+   fun:_rt_boot
+}

[Valgrind-developers] [valgrind] Solaris regtest: add to fdleak filter for file_dclose

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=930a17c6cbe27fa923c1b22ee1deae9875614b5f

commit 930a17c6cbe27fa923c1b22ee1deae9875614b5f
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 14 08:15:00 2025 +0100

    Solaris regtest: add to fdleak filter for file_dclose

Diff:
---
 none/tests/filter_fdleak | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/none/tests/filter_fdleak b/none/tests/filter_fdleak
index 6ad1ac6f11..260c6e24c0 100755
--- a/none/tests/filter_fdleak
+++ b/none/tests/filter_fdleak
@@ -34,6 +34,8 @@ perl -p -e 's/ __write / write /' |
 sed '/ _so_socket /d;' |
 awk '/ socket /{sub(/ by /, " at "); print; next}{print}' |
 awk '/ at .* _syscall6 /{getline; getline; sub(/ by /, " at "); print; next}{print}' |
+sed '/__open/d' |
+sed 's/__systemcall6/creat/' |
 
 # Stack traces showing where fds were created have lots of variations:
 # different numbers of entries, line numbers or not depending on whether

[Valgrind-developers] [valgrind] Revert "FreeBSD regtest: Implement setaffinity in gdb sleepers for nlvgdbsigqueue"

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=79d09d0821ed55aec8ddf07c1207c9e290d1b1c9

commit 79d09d0821ed55aec8ddf07c1207c9e290d1b1c9
Author: Paul Floyd <pj...@wa...>
Date:   Sun Feb 9 08:21:58 2025 +0100

    Revert "FreeBSD regtest: Implement setaffinity in gdb sleepers for nlvgdbsigqueue"
    
    This reverts commit 98cc2209b5c1690a8d181d372c6a248fbd3b2134.

Diff:
---
 gdbserver_tests/sleepers.c | 10 ++--------
 1 file changed, 2 insertions(+), 8 deletions(-)

diff --git a/gdbserver_tests/sleepers.c b/gdbserver_tests/sleepers.c
index e884c9b8fe..b433c19847 100644
--- a/gdbserver_tests/sleepers.c
+++ b/gdbserver_tests/sleepers.c
@@ -15,7 +15,7 @@ static int sleepms = 1000; // in each loop, will sleep "sleepms" milliseconds
 static int burn = 0; // after each sleep, will burn cpu in a tight 'burn' loop 
 static void setup_sigusr_handler(void); // sigusr1 and 2 sigaction setup.
 
-static pid_t gettid_sys(void)
+static pid_t gettid_sys()
 {
 #ifdef __NR_gettid
    return syscall(__NR_gettid);
@@ -32,7 +32,7 @@ void whoami(char *msg)
 }
 
 
-static void do_burn(void)
+static void do_burn ()
 {
    int i;
    int loopnr = 0;
@@ -126,12 +126,6 @@ static void setaffinity(void)
    CPU_ZERO(&single_cpu);
    CPU_SET(1, &single_cpu);
    (void) sched_setaffinity(0, sizeof(single_cpu), &single_cpu);
-#endif
-#ifdef VGO_freebsd
-   cpu_set_t single_cpu;
-   CPU_ZERO(&single_cpu);
-   CPU_SET(1, &single_cpu);
-   pthread_setaffinity_np(pthread_self(), sizeof(single_cpu), &single_cpu);
 #endif
    // GDBTD: equivalent for Darwin ?
 }

[Valgrind-developers] [valgrind] FreeBSD regtest: a couple of updates for FreeBSD 13.5 x86

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=f2bb8c345f4e35deae99b904e77d23678744037b

commit f2bb8c345f4e35deae99b904e77d23678744037b
Author: Paul Floyd <pj...@wa...>
Date:   Sat Feb 8 21:47:38 2025 +0100

    FreeBSD regtest: a couple of updates for FreeBSD 13.5 x86

Diff:
---
 none/tests/file_dclose.supp | 9 ++++++++-
 none/tests/filter_fdleak    | 1 +
 2 files changed, 9 insertions(+), 1 deletion(-)

diff --git a/none/tests/file_dclose.supp b/none/tests/file_dclose.supp
index d812adeada..7c8bacae1f 100644
--- a/none/tests/file_dclose.supp
+++ b/none/tests/file_dclose.supp
@@ -3,5 +3,12 @@
    CoreError:FdBadClose
    fun:*close*
    fun:closefile
-   #fun:main
 }
+{
+   badclose_2
+   CoreError:FdBadClose
+   fun:_close
+   fun:close
+   fun:closefile
+}
+
diff --git a/none/tests/filter_fdleak b/none/tests/filter_fdleak
index 1f4a8c7bd6..6ad1ac6f11 100755
--- a/none/tests/filter_fdleak
+++ b/none/tests/filter_fdleak
@@ -26,6 +26,7 @@ perl -p -e "s/write\.c:[1-9][0-9]*/in \/...libc.../" |
 perl -p -e 's/ _close / close /;s/ _openat / creat /;s/ _write/ write/;s/internet/AF_INET socket 4: 127.0.0.1:... <-> 127.0.0.1:.../' |
 sed '/by 0x........: close (in \/...libc...)/d' |
 sed '/by 0x........: creat (in \/...libc...)/d' |
+sed '/by 0x........: write (in \/...libc...)/d' |
 
 # Solaris specific fdleak filters
 perl -p -e 's/ __close / close /' |

[Valgrind-developers] [valgrind] Linux regtest: broaden none/tests/linux/bug498317.supp

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=44e0c761af15114c42c76d2102b7fe01358de2f6

commit 44e0c761af15114c42c76d2102b7fe01358de2f6
Author: Paul Floyd <pj...@wa...>
Date:   Sat Feb 8 10:35:38 2025 +0100

    Linux regtest: broaden none/tests/linux/bug498317.supp
    
    fcntl function name is versioned on Raspberry Pi OS
    (fcntl@@GLIBC_2.28) so use a wildcard in the suppression.

Diff:
---
 none/tests/linux/bug498317.supp | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/none/tests/linux/bug498317.supp b/none/tests/linux/bug498317.supp
index f345f3771f..4e204304f0 100644
--- a/none/tests/linux/bug498317.supp
+++ b/none/tests/linux/bug498317.supp
@@ -2,7 +2,7 @@
    test suppression of FdBadUse
    CoreError:FdBadUse
    ...
-   fun:fcntl
+   fun:fcntl*
    fun:main
 }

[Valgrind-developers] [valgrind] FreeBSD regtest: Implement setaffinity in gdb sleepers for nlvgdbsigqueue

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=98cc2209b5c1690a8d181d372c6a248fbd3b2134

commit 98cc2209b5c1690a8d181d372c6a248fbd3b2134
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 7 21:29:23 2025 +0100

    FreeBSD regtest: Implement setaffinity in gdb sleepers for nlvgdbsigqueue
    
    Doesn't seem to help, the test is still very flaky on recent (2025) hardware.

Diff:
---
 gdbserver_tests/sleepers.c | 10 ++++++++--
 1 file changed, 8 insertions(+), 2 deletions(-)

diff --git a/gdbserver_tests/sleepers.c b/gdbserver_tests/sleepers.c
index b433c19847..e884c9b8fe 100644
--- a/gdbserver_tests/sleepers.c
+++ b/gdbserver_tests/sleepers.c
@@ -15,7 +15,7 @@ static int sleepms = 1000; // in each loop, will sleep "sleepms" milliseconds
 static int burn = 0; // after each sleep, will burn cpu in a tight 'burn' loop 
 static void setup_sigusr_handler(void); // sigusr1 and 2 sigaction setup.
 
-static pid_t gettid_sys()
+static pid_t gettid_sys(void)
 {
 #ifdef __NR_gettid
    return syscall(__NR_gettid);
@@ -32,7 +32,7 @@ void whoami(char *msg)
 }
 
 
-static void do_burn ()
+static void do_burn(void)
 {
    int i;
    int loopnr = 0;
@@ -126,6 +126,12 @@ static void setaffinity(void)
    CPU_ZERO(&single_cpu);
    CPU_SET(1, &single_cpu);
    (void) sched_setaffinity(0, sizeof(single_cpu), &single_cpu);
+#endif
+#ifdef VGO_freebsd
+   cpu_set_t single_cpu;
+   CPU_ZERO(&single_cpu);
+   CPU_SET(1, &single_cpu);
+   pthread_setaffinity_np(pthread_self(), sizeof(single_cpu), &single_cpu);
 #endif
    // GDBTD: equivalent for Darwin ?
 }

[Valgrind-developers] [valgrind] Remove exec bits on a few files and delete a Solaris regtest binary

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=b918889d6cf18243df8e30a3462efa093b6372ac

commit b918889d6cf18243df8e30a3462efa093b6372ac
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 7 12:37:04 2025 +0100

    Remove exec bits on a few files and delete a Solaris regtest binary
    
    Pushed the binary by mistake in Sept 2023!

Diff:
---
 VEX/priv/guest_nanomips_toIR.c                     |   0
 VEX/priv/host_nanomips_defs.c                      |   0
 drd/tests/pth_detached3.c                          |   0
 memcheck/tests/sized_aligned_new_delete_misaligned | Bin 42384 -> 0 bytes
 4 files changed, 0 insertions(+), 0 deletions(-)

diff --git a/VEX/priv/guest_nanomips_toIR.c b/VEX/priv/guest_nanomips_toIR.c
old mode 100755
new mode 100644
diff --git a/VEX/priv/host_nanomips_defs.c b/VEX/priv/host_nanomips_defs.c
old mode 100755
new mode 100644
diff --git a/drd/tests/pth_detached3.c b/drd/tests/pth_detached3.c
old mode 100755
new mode 100644
diff --git a/memcheck/tests/sized_aligned_new_delete_misaligned b/memcheck/tests/sized_aligned_new_delete_misaligned
deleted file mode 100755
index 1c69036c16..0000000000
Binary files a/memcheck/tests/sized_aligned_new_delete_misaligned and /dev/null differ

[Valgrind-developers] [valgrind] FreeBSD regtest: add arm64 expected for auxv_script test

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=410ff526ea55f22cb4f301fde1a372046ab2149e

commit 410ff526ea55f22cb4f301fde1a372046ab2149e
Author: Paul Floyd <pj...@wa...>
Date:   Fri Feb 7 08:38:38 2025 +0100

    FreeBSD regtest: add arm64 expected for auxv_script test

Diff:
---
 none/tests/freebsd/Makefile.am                  |  1 +
 none/tests/freebsd/auxv_script.stderr.exp-arm64 | 30 +++++++++++++++++++++++++
 2 files changed, 31 insertions(+)

diff --git a/none/tests/freebsd/Makefile.am b/none/tests/freebsd/Makefile.am
index 667445aa15..15ea01cc9e 100644
--- a/none/tests/freebsd/Makefile.am
+++ b/none/tests/freebsd/Makefile.am
@@ -13,6 +13,7 @@ EXTRA_DIST = \
 	auxv.stderr.exp-arm64 \
 	auxv_script.vgtest \
 	auxv_script.stderr.exp \
+	auxv_script.stderr.exp-arm64 \
 	bug452274.vgtest \
 	bug452274.stderr.exp \
 	bug498317.vgtest bug498317.stderr.exp \
diff --git a/none/tests/freebsd/auxv_script.stderr.exp-arm64 b/none/tests/freebsd/auxv_script.stderr.exp-arm64
new file mode 100644
index 0000000000..738a284dcb
--- /dev/null
+++ b/none/tests/freebsd/auxv_script.stderr.exp-arm64
@@ -0,0 +1,30 @@
+val: AT_PHDR int: 03 ptr: 0x........
+val: AT_PHENT int: 04 ptr: 0x........
+val: AT_PHNUM int: 05 ptr: 0x........
+val: AT_PAGESZ int: 06 ptr: 0x........
+val: AT_FLAGS int: 08 ptr: 0x........
+val: AT_ENTRY int: 09 ptr: 0x........
+val: AT_BASE int: 07 ptr: 0x........
+val: AT_EHDRFLAGS int: 24 ptr: 0x........
+val: AT_EXECPATH int: 15 ptr: 0x........
+EXECPATH: BASEDIR/valgrind/none/tests/freebsd/auxv
+val: AT_OSRELDATE int: 18 ptr: 0x........
+val: AT_CANARY int: 16 ptr: 0x........
+val: AT_CANARYLEN int: 17 ptr: 0x........
+val: AT_NCPUS int: 19 ptr: 0x........
+val: AT_PAGESIZES int: 20 ptr: 0x........
+val: AT_PAGESIZESLEN int: 21 ptr: 0x........
+val: AT_IGNORE int: 01 ptr: 0x........
+val: AT_STACKPROT int: 23 ptr: 0x........
+val: AT_HWCAP int: 25 ptr: 0x........
+val: AT_IGNORE int: 01 ptr: 0x........
+val: AT_IGNORE int: 01 ptr: 0x........
+val: AT_ARGC int: 28 ptr: 0x........
+val: AT_ARGV int: 29 ptr: 0x........
+ARGV: ./auxv
+val: AT_ENVC int: 30 ptr: 0x........
+val: AT_ENVV int: 31 ptr: 0x........
+val: AT_PS_STRINGS int: 32 ptr: 0x........
+PS_STRINGS ARGV: ./auxv
+val: AT_USRSTACKBASE int: 35 ptr: 0x........
+val: AT_USRSTACKLIM int: 36 ptr: 0x........

[Valgrind-developers] [valgrind] FreeBSD DRD: cleanup suppressions

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=fc9ae97c3d3b93a021328c029503070d2acdc344

commit fc9ae97c3d3b93a021328c029503070d2acdc344
Author: Paul Floyd <pj...@wa...>
Date:   Thu Feb 6 21:49:59 2025 +0100

    FreeBSD DRD: cleanup suppressions
    
    There were quite a lot that duplicated the top level libthr.so.3 suppression.

Diff:
---
 freebsd-drd.supp | 47 +----------------------------------------------
 1 file changed, 1 insertion(+), 46 deletions(-)

diff --git a/freebsd-drd.supp b/freebsd-drd.supp
index 2f3c8066a9..c0767aadf7 100644
--- a/freebsd-drd.supp
+++ b/freebsd-drd.supp
@@ -23,28 +23,7 @@
    fun:f*lockfile
 }
 {
-   DRD-MANY1
-   drd:ConflictingAccess
-   obj:*/lib*/libthr.so.3
-   obj:/libexec/ld-elf*.so.1
-}
-{
-   DRD-MANY2
-   drd:ConflictingAccess
-   obj:*/lib*/libthr.so.3
-   obj:*/lib*/libthr.so.3
-   obj:/libexec/ld-elf*.so.1
-}
-{
-   DRD-MANY3
-   drd:ConflictingAccess
-   obj:*/lib*/libthr.so.3
-   obj:*/lib*/libthr.so.3
-   obj:*/lib*/libthr.so.3
-   obj:/libexec/ld-elf*.so.1
-}
-{
-   DRD-MANY4
+   DRD-MANY
    drd:ConflictingAccess
    obj:*/lib*/libthr.so.3
 }
@@ -75,14 +54,6 @@
    drd:ConflictingAccess
    fun:.omp_outlined*
 }
-{
-   DRD-LIBGOMP4
-   drd:ConflictingAccess
-   obj:*/lib*/libthr.so.3
-   obj:*/lib*/libthr.so.3
-   obj:*/lib*/libthr.so.3
-   fun:pthread_setspecific
-}
 {
    DRD-UNWIND1
    drd:ConflictingAccess
@@ -135,22 +106,6 @@
    obj:*/lib*/libgcc_s.so.1
    obj:*/lib*/libgcc_s.so.1
 }
-{
-   DRD-DLOPEN
-   drd:ConflictingAccess
-   obj:*/lib*/libthr.so.3
-   obj:/libexec/ld-elf*.so.1
-   obj:/libexec/ld-elf*.so.1
-   obj:/libexec/ld-elf*.so.1
-   fun:dlopen
-}
-{
-   DRD-LD-ELF
-   drd:ConflictingAccess
-   obj:*/lib*/libthr.so.3
-   ...
-   obj:/libexec/ld-elf*.so.1
-}
 {
    DRD-SNPRINTF
    drd:ConflictingAccess

[Valgrind-developers] [valgrind] Bug 396415 - Valgrind is not looking up $ORIGIN rpath of shebang programs

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=1fefba79021779d840bbf8cebc43e40c74b40f31

commit 1fefba79021779d840bbf8cebc43e40c74b40f31
Author: Paul Floyd <pj...@wa...>
Date:   Thu Feb 6 20:23:42 2025 +0100

    Bug 396415 - Valgrind is not looking up $ORIGIN rpath of shebang programs

Diff:
---
 NEWS                                      |  1 +
 coregrind/m_initimg/initimg-darwin.c      |  7 ----
 coregrind/m_initimg/initimg-freebsd.c     | 59 ++++++++++++++++++++++++++-----
 coregrind/m_initimg/initimg-linux.c       |  7 ----
 coregrind/m_initimg/initimg-solaris.c     |  7 ----
 coregrind/m_ume/elf.c                     |  7 ++++
 coregrind/m_ume/macho.c                   |  5 +++
 none/tests/freebsd/Makefile.am            |  2 ++
 none/tests/freebsd/auxv_script            |  1 +
 none/tests/freebsd/auxv_script.stderr.exp | 29 +++++++++++++++
 none/tests/freebsd/auxv_script.vgtest     |  4 +++
 11 files changed, 100 insertions(+), 29 deletions(-)

diff --git a/NEWS b/NEWS
index ca5bfb23f4..1c627c7744 100644
--- a/NEWS
+++ b/NEWS
@@ -32,6 +32,7 @@ bugzilla (https://bugs.kde.org/enter_bug.cgi?product=valgrind) rather
 than mailing the developers (or mailing lists) directly -- bugs that
 are not entered into bugzilla tend to get forgotten about or ignored.
 
+396415  Valgrind is not looking up $ORIGIN rpath of shebang programs
 487296  --track-fds=yes and --track-fds=all report erroneous information
         when fds 0, 1, or 2 are used as non-std
 489913  WARNING: unhandled amd64-linux syscall: 444 (landlock_create_ruleset)
diff --git a/coregrind/m_initimg/initimg-darwin.c b/coregrind/m_initimg/initimg-darwin.c
index 7bac7dc0d1..57fbd0bf75 100644
--- a/coregrind/m_initimg/initimg-darwin.c
+++ b/coregrind/m_initimg/initimg-darwin.c
@@ -61,7 +61,6 @@ static void load_client ( /*OUT*/ExeInfo* info,
 {
    const HChar* exe_name;
    Int    ret;
-   SysRes res;
 
    vg_assert( VG_(args_the_exename) != NULL);
    exe_name = VG_(find_executable)( VG_(args_the_exename) );
@@ -76,12 +75,6 @@ static void load_client ( /*OUT*/ExeInfo* info,
 
    // The client was successfully loaded!  Continue.
 
-   /* Get hold of a file descriptor which refers to the client
-      executable.  This is needed for attaching to GDB. */
-   res = VG_(open)(exe_name, VKI_O_RDONLY, VKI_S_IRUSR);
-   if (!sr_isError(res))
-      VG_(cl_exec_fd) = sr_Res(res);
-
    /* Copy necessary bits of 'info' that were filled in */
    *client_ip  = info->init_ip;
 }
diff --git a/coregrind/m_initimg/initimg-freebsd.c b/coregrind/m_initimg/initimg-freebsd.c
index 6749e849fc..cb98eb9144 100644
--- a/coregrind/m_initimg/initimg-freebsd.c
+++ b/coregrind/m_initimg/initimg-freebsd.c
@@ -64,7 +64,6 @@ static void load_client ( /*OUT*/ExeInfo* info,
 {
    const HChar* exe_name;
    Int    ret;
-   SysRes res;
 
    vg_assert( VG_(args_the_exename) != NULL);
    exe_name = VG_(find_executable)( VG_(args_the_exename) );
@@ -83,13 +82,6 @@ static void load_client ( /*OUT*/ExeInfo* info,
 
    // The client was successfully loaded!  Continue.
 
-   /* Get hold of a file descriptor which refers to the client
-      executable.  This is needed for attaching to GDB. */
-   res = VG_(open)(exe_name, VKI_O_RDONLY, VKI_S_IRUSR);
-   if (!sr_isError(res)) {
-      VG_(cl_exec_fd) = sr_Res(res);
-   }
-
    /* Copy necessary bits of 'info' that were filled in */
    *client_ip  = info->init_ip;
    *client_toc = info->init_toc;
@@ -342,6 +334,53 @@ static const struct auxv *find_auxv(const UWord* sp)
    return (const struct auxv *)sp;
 }
 
+static Bool try_get_interp(const HChar* args_exe, HChar* interp_out)
+{
+   HChar  hdr[4096];
+   Int    len = sizeof hdr;
+   SysRes res;
+   Int fd;
+   HChar* end;
+   HChar* cp;
+   HChar* interp;
+
+   res = VG_(open)(args_exe, VKI_O_RDONLY, 0);
+   if (sr_isError(res)) {
+      return False;
+   } else {
+      fd = sr_Res(res);
+   }
+
+   res = VG_(pread)(fd, hdr, len, 0);
+
+   if (sr_isError(res)) {
+      VG_(close)(fd);
+      return False;
+   } else {
+      len = sr_Res(res);
+   }
+
+   if (0 != VG_(memcmp)(hdr, "#!", 2)) {
+       VG_(close)(fd);
+      return False;
+   }
+
+   end    = hdr + len;
+   interp = hdr + 2;
+   while (interp < end && (*interp == ' ' || *interp == '\t'))
+      interp++;
+
+   for (cp = interp; cp < end && !VG_(isspace)(*cp); cp++)
+      ;
+
+   *cp = '\0';
+
+   VG_(sprintf)(interp_out, "%s", interp);
+
+   VG_(close)(fd);
+   return True;
+}
+
 /* ----------------------------------------------------------------
 
    This sets up the client's initial stack, containing the args,
@@ -425,6 +464,10 @@ static Addr setup_client_stack(const void*  init_sp,
    vg_assert( VG_(args_for_client) );
 
    const HChar *exe_name = VG_(find_executable)(VG_(args_the_exename));
+   HChar interp_name[VKI_PATH_MAX];
+   if (try_get_interp(exe_name, interp_name)) {
+      exe_name = interp_name;
+   }
    HChar resolved_name[VKI_PATH_MAX];
    VG_(realpath)(exe_name, resolved_name);
 
diff --git a/coregrind/m_initimg/initimg-linux.c b/coregrind/m_initimg/initimg-linux.c
index 5359189bf6..feec1e5f11 100644
--- a/coregrind/m_initimg/initimg-linux.c
+++ b/coregrind/m_initimg/initimg-linux.c
@@ -70,7 +70,6 @@ static void load_client ( /*MOD*/ExeInfo* info,
 {
    const HChar* exe_name;
    Int    ret;
-   SysRes res;
 
    vg_assert( VG_(args_the_exename) != NULL);
    exe_name = VG_(find_executable)( VG_(args_the_exename) );
@@ -88,12 +87,6 @@ static void load_client ( /*MOD*/ExeInfo* info,
 
    // The client was successfully loaded!  Continue.
 
-   /* Get hold of a file descriptor which refers to the client
-      executable.  This is needed for attaching to GDB. */
-   res = VG_(open)(exe_name, VKI_O_RDONLY, VKI_S_IRUSR);
-   if (!sr_isError(res))
-      VG_(cl_exec_fd) = sr_Res(res);
-
    /* Copy necessary bits of 'info' that were filled in */
    *client_ip  = info->init_ip;
    *client_toc = info->init_toc;
diff --git a/coregrind/m_initimg/initimg-solaris.c b/coregrind/m_initimg/initimg-solaris.c
index 1e9e3fd061..4fa94b8c4c 100644
--- a/coregrind/m_initimg/initimg-solaris.c
+++ b/coregrind/m_initimg/initimg-solaris.c
@@ -65,7 +65,6 @@ static void load_client(/*OUT*/ExeInfo *info,
 {
    const HChar *exe_name;
    Int ret;
-   SysRes res;
 
    vg_assert(VG_(args_the_exename));
    exe_name = VG_(find_executable)(VG_(args_the_exename));
@@ -96,12 +95,6 @@ static void load_client(/*OUT*/ExeInfo *info,
    }
    VG_(strcpy)(out_exe_name, exe_name);
 
-   /* Get hold of a file descriptor which refers to the client executable.
-      This is needed for attaching to GDB. */
-   res = VG_(open)(exe_name, VKI_O_RDONLY, VKI_S_IRUSR);
-   if (!sr_isError(res))
-      VG_(cl_exec_fd) = sr_Res(res);
-
    /* Set initial brk values. */
    if (info->ldsoexec) {
       VG_(brk_base) = VG_(brk_limit) = -1;
diff --git a/coregrind/m_ume/elf.c b/coregrind/m_ume/elf.c
index a850a50cfa..bad7e5ee6b 100644
--- a/coregrind/m_ume/elf.c
+++ b/coregrind/m_ume/elf.c
@@ -41,6 +41,7 @@
 #include "pub_core_mallocfree.h"    // VG_(malloc), VG_(free)
 #include "pub_core_vkiscnums.h"
 #include "pub_core_syscall.h"       // VG_(strerror)
+#include "pub_core_clientstate.h"
 #include "pub_core_ume.h"           // self
 
 #include "priv_ume.h"
@@ -876,6 +877,12 @@ Int VG_(load_ELF)(Int fd, const HChar* name, /*MOD*/ExeInfo* info)
    VG_(free)(e->p);
    VG_(free)(e);
 
+   /* Get hold of a file descriptor which refers to the client
+      executable.  This is needed for attaching to GDB. */
+   SysRes res = VG_(dup)(fd);
+   if (!sr_isError(res))
+      VG_(cl_exec_fd) = sr_Res(res);
+
    return 0;
 }
 
diff --git a/coregrind/m_ume/macho.c b/coregrind/m_ume/macho.c
index 046d174933..6d65574e44 100644
--- a/coregrind/m_ume/macho.c
+++ b/coregrind/m_ume/macho.c
@@ -41,6 +41,7 @@
 #include "pub_core_machine.h"       // VG_ELF_CLASS (XXX: which should be moved)
 #include "pub_core_mallocfree.h"    // VG_(malloc), VG_(free)
 #include "pub_core_syscall.h"       // VG_(strerror)
+#include "pub_core_clientstate.h"
 #include "pub_core_ume.h"           // self
 
 #include "priv_ume.h"
@@ -857,6 +858,10 @@ Int VG_(load_macho)(Int fd, const HChar *name, ExeInfo *info)
 
    info->executable_path = VG_(strdup)("ume.macho.executable_path", name);
 
+   SysRes res = VG_(dup)(fd);
+   if (!sr_isError(res))
+      VG_(cl_exec_fd) = sr_Res(res);
+
    return 0;
 }
 
diff --git a/none/tests/freebsd/Makefile.am b/none/tests/freebsd/Makefile.am
index d1f7759ec9..667445aa15 100644
--- a/none/tests/freebsd/Makefile.am
+++ b/none/tests/freebsd/Makefile.am
@@ -11,6 +11,8 @@ EXTRA_DIST = \
 	auxv.stderr.exp-freebsd131 \
 	auxv.stderr.exp-freebsd14 \
 	auxv.stderr.exp-arm64 \
+	auxv_script.vgtest \
+	auxv_script.stderr.exp \
 	bug452274.vgtest \
 	bug452274.stderr.exp \
 	bug498317.vgtest bug498317.stderr.exp \
diff --git a/none/tests/freebsd/auxv_script b/none/tests/freebsd/auxv_script
new file mode 100755
index 0000000000..7794012c1f
--- /dev/null
+++ b/none/tests/freebsd/auxv_script
@@ -0,0 +1 @@
+#!./auxv
diff --git a/none/tests/freebsd/auxv_script.stderr.exp b/none/tests/freebsd/auxv_script.stderr.exp
new file mode 100644
index 0000000000..dcca09f63c
--- /dev/null
+++ b/none/tests/freebsd/auxv_script.stderr.exp
@@ -0,0 +1,29 @@
+val: AT_PHDR int: 03 ptr: 0x........
+val: AT_PHENT int: 04 ptr: 0x........
+val: AT_PHNUM int: 05 ptr: 0x........
+val: AT_PAGESZ int: 06 ptr: 0x........
+val: AT_FLAGS int: 08 ptr: 0x........
+val: AT_ENTRY int: 09 ptr: 0x........
+val: AT_BASE int: 07 ptr: 0x........
+val: AT_EHDRFLAGS int: 24 ptr: 0x........
+val: AT_EXECPATH int: 15 ptr: 0x........
+EXECPATH: BASEDIR/valgrind/none/tests/freebsd/auxv
+val: AT_OSRELDATE int: 18 ptr: 0x........
+val: AT_CANARY int: 16 ptr: 0x........
+val: AT_CANARYLEN int: 17 ptr: 0x........
+val: AT_NCPUS int: 19 ptr: 0x........
+val: AT_PAGESIZES int: 20 ptr: 0x........
+val: AT_PAGESIZESLEN int: 21 ptr: 0x........
+val: AT_IGNORE int: 01 ptr: 0x........
+val: AT_STACKPROT int: 23 ptr: 0x........
+val: AT_IGNORE int: 01 ptr: 0x........
+val: AT_ARGC int: 28 ptr: 0x........
+val: AT_ARGV int: 29 ptr: 0x........
+ARGV: ./auxv
+val: AT_ENVC int: 30 ptr: 0x........
+val: AT_ENVV int: 31 ptr: 0x........
+val: AT_PS_STRINGS int: 32 ptr: 0x........
+PS_STRINGS ARGV: ./auxv
+val: AT_IGNORE int: 01 ptr: 0x........
+val: AT_USRSTACKBASE int: 35 ptr: 0x........
+val: AT_USRSTACKLIM int: 36 ptr: 0x........
diff --git a/none/tests/freebsd/auxv_script.vgtest b/none/tests/freebsd/auxv_script.vgtest
new file mode 100644
index 0000000000..ccd450a9a2
--- /dev/null
+++ b/none/tests/freebsd/auxv_script.vgtest
@@ -0,0 +1,4 @@
+prog: auxv_script
+vgopts: -q
+stderr_filter: filter_auxv
+

[Valgrind-developers] Regarding the betterment of the LoongArch64/Linux port

[Binbin Z. <zho...@lo...>]

Hi Mark, Paul & all developers:

First, it is a pleasure to meet you all, and I am honored to 
continue LoongArch64/Linux's port work.

Regarding the Valgrind LoongAcrh support, we are continuously working on it.
Specifically, the LoongAcrh patch set was previously updated to v8[1], 
while we continue to test on the latest Valgrind version.
Additionally, I'm also continuing to work on improving vector 
instruction support.

Unfortunately, probably due to a lack of understanding of the community 
support process for the new architecture, port work seems to go on for a 
long time, so what do we need to pay particular attention to?

I'm not sure if the **CI build** of the new architecture is a critical 
step. If it is, I would try to provide an available environment.

There are two ways to provide a CI build environment, which one does our 
community need:
1. Donate machines directly to our community;
2. Provide a community-accessible machine environment.

[1]: https://bugs.kde.org/show_bug.cgi?id=457504

Thanks.
Binbin

[Valgrind-developers] [valgrind] FreeBSD syscall: improve utrace wrapper

[Paul F. <pa...@so...>]

https://sourceware.org/git/gitweb.cgi?p=valgrind.git;h=6f6025d53a8f53a85c74d03d7b177792f62ac147

commit 6f6025d53a8f53a85c74d03d7b177792f62ac147
Author: Paul Floyd <pj...@wa...>
Date:   Tue Feb 4 20:34:57 2025 +0100

    FreeBSD syscall: improve utrace wrapper
    
    The public wrapper is just a void pointer and length.
    Internally, the struct used has holes which generate
    errors (since ld.so which is probably the only user
    of this syscall doesn't memset the struct to zero). Added
    the struct definition and PRE_MEM_READ for each field.

Diff:
---
 .gitignore                                   |  1 +
 coregrind/m_syswrap/syswrap-freebsd.c        | 29 +++++++++++++++++++++++++++-
 memcheck/tests/freebsd/Makefile.am           |  5 ++++-
 memcheck/tests/freebsd/scalar.stderr.exp     |  4 ----
 memcheck/tests/freebsd/scalar.stderr.exp-x86 |  4 ----
 memcheck/tests/freebsd/utrace.stderr.exp     |  0
 memcheck/tests/freebsd/utrace.vgtest         |  6 ++++++
 7 files changed, 39 insertions(+), 10 deletions(-)

diff --git a/.gitignore b/.gitignore
index 483b077ca3..c5a2b0592c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1452,6 +1452,7 @@
 /memcheck/tests/freebsd/timing_safe
 /memcheck/tests/freebsd/utimens
 /memcheck/tests/freebsd/utimes
+/memcheck/tests/freebsd/utrace
 
 # /memcheck/tests/amd64-freebsd
 /memcheck/tests/amd64-freebsd/*.stderr.diff
diff --git a/coregrind/m_syswrap/syswrap-freebsd.c b/coregrind/m_syswrap/syswrap-freebsd.c
index 3397249383..6fd4367a83 100644
--- a/coregrind/m_syswrap/syswrap-freebsd.c
+++ b/coregrind/m_syswrap/syswrap-freebsd.c
@@ -3109,13 +3109,40 @@ POST(sys_sched_rr_get_interval)
    POST_MEM_WRITE(ARG2, sizeof(struct vki_timespec));
 }
 
+/*
+ * Putting this here rather than in vki-freebsd.h because this is a workaround
+ * (see https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=284563)
+ * The syscall interface doesn't allow us to properly validate the memory,
+ * and struct utrace_rtld isn't public.
+ */
+#define	VKI_RTLD_UTRACE_SIG_SZ 4
+
+struct vki_utrace_rtld {
+    char sig[VKI_RTLD_UTRACE_SIG_SZ];
+    int event;
+    void *handle;
+    void *mapbase;
+    size_t mapsize;
+    int refcnt;
+    char name[VKI_PATH_MAX];
+};
+
 // SYS_utrace  335
 // int utrace(const void *addr, size_t len);
 PRE(sys_utrace)
 {
    PRINT("sys_utrace ( %#" FMT_REGWORD "x, %" FMT_REGWORD "u )", ARG1, ARG2);
    PRE_REG_READ2(int, "utrace", const void *, addr, vki_size_t, len);
-   PRE_MEM_READ( "utrace(addr)", ARG1, ARG2 );
+   if (ARG1 && ARG2 >= sizeof(struct vki_utrace_rtld) && ML_(safe_to_deref)((const void*)ARG1, ARG2)) {
+       struct vki_utrace_rtld* ut = (struct vki_utrace_rtld*)ARG1;
+       PRE_MEM_READ("utrace(addr.sig)", (Addr)&ut->sig, VKI_RTLD_UTRACE_SIG_SZ*sizeof(char));
+       PRE_MEM_READ("utrace(addr.event)", (Addr)&ut->event, sizeof(int));
+       PRE_MEM_READ("utrace(addr.handle)", (Addr)&ut->handle, sizeof(void*));
+       PRE_MEM_READ("utrace(addr.mapbase)", (Addr)&ut->mapbase, sizeof(void*));
+       PRE_MEM_READ("utrace(addr.mapsize)", (Addr)&ut->mapsize, sizeof(size_t));
+       PRE_MEM_READ("utrace(addr.refcnt)", (Addr)&ut->handle, sizeof(int));
+       PRE_MEM_READ("utrace(addr.name)", (Addr)&ut->name, VKI_PATH_MAX*sizeof(char));
+   }
 }
 
 // SYS_kldsym  337
diff --git a/memcheck/tests/freebsd/Makefile.am b/memcheck/tests/freebsd/Makefile.am
index 4b6b36c6ce..1213b31898 100644
--- a/memcheck/tests/freebsd/Makefile.am
+++ b/memcheck/tests/freebsd/Makefile.am
@@ -137,7 +137,10 @@ EXTRA_DIST = \
 	utimens.stderr.exp \
 	utimes.vgtest \
 	utimes.stderr.exp-x86 \
-	utimes.stderr.exp
+	utimes.stderr.exp \
+	utrace.vgtest \
+	utrace.stderr.exp
+
 
 check_PROGRAMS = \
 	access aio aio_read aligned_alloc bug464476 bug470713 \
diff --git a/memcheck/tests/freebsd/scalar.stderr.exp b/memcheck/tests/freebsd/scalar.stderr.exp
index 4a57e002aa..0e47fe1aac 100644
--- a/memcheck/tests/freebsd/scalar.stderr.exp
+++ b/memcheck/tests/freebsd/scalar.stderr.exp
@@ -2391,10 +2391,6 @@ Syscall param utrace(addr) contains uninitialised byte(s)
 Syscall param utrace(len) contains uninitialised byte(s)
    ...
 
-Syscall param utrace(addr) points to unaddressable byte(s)
-   ...
- Address 0x........ is not stack'd, malloc'd or (recently) free'd
-
 ---------------------------------------------------------
 337:              SYS_kldsym 3s 1m
 ---------------------------------------------------------
diff --git a/memcheck/tests/freebsd/scalar.stderr.exp-x86 b/memcheck/tests/freebsd/scalar.stderr.exp-x86
index 86599836de..6a532ae8e0 100644
--- a/memcheck/tests/freebsd/scalar.stderr.exp-x86
+++ b/memcheck/tests/freebsd/scalar.stderr.exp-x86
@@ -2397,10 +2397,6 @@ Syscall param utrace(addr) contains uninitialised byte(s)
 Syscall param utrace(len) contains uninitialised byte(s)
    ...
 
-Syscall param utrace(addr) points to unaddressable byte(s)
-   ...
- Address 0x........ is not stack'd, malloc'd or (recently) free'd
-
 ---------------------------------------------------------
 337:              SYS_kldsym 3s 1m
 ---------------------------------------------------------
diff --git a/memcheck/tests/freebsd/utrace.stderr.exp b/memcheck/tests/freebsd/utrace.stderr.exp
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/memcheck/tests/freebsd/utrace.vgtest b/memcheck/tests/freebsd/utrace.vgtest
new file mode 100644
index 0000000000..2b69572982
--- /dev/null
+++ b/memcheck/tests/freebsd/utrace.vgtest
@@ -0,0 +1,6 @@
+# This produces no output (only does so under truss or ktrace)
+# but there was an issue with the syscall that generated false positives
+# due to holes in the struct used
+env: LD_UTRACE=1
+prog: ../../../tests/true
+vgopts: -q

Re: [Valgrind-developers] Request to contribute on SSE4a support

[Paul F. <pj...@wa...>]

On 01-02-25 00:34, Pol Marcet Sardà wrote:
> I have attached the patch in: https://bugs.kde.org/show_bug.cgi? 
> id=486398 <https://bugs.kde.org/show_bug.cgi?id=486398>
> If someone can review/submit the patch, they can freely do so!

I'll take a look as soon as I can.

A+
Paul

Re: [Valgrind-developers] Request to contribute on SSE4a support

[Pol M. S. <pol...@gm...>]

I have attached the patch in: https://bugs.kde.org/show_bug.cgi?id=486398
If someone can review/submit the patch, they can freely do so!

Missatge de Pol Marcet Sardà <pol...@gm...> del dia dv., 31 de
gen. 2025 a les 23:54:

> Hello,
>
> I'd need some assistance on how to submit a patch to the sourceware repo,
> at the moment I have a patch cleaned up with all instructions implemented.
> I am unfamiliar with the process (I am used to the PR mechanisms of
> bitbucket/gitlab/github)
>
> Sorry for writing a similar message to both the mailing list and the IRC
>
> Best regards,
>
> Pol Marcet Sardà
>
> Missatge de Pol Marcet Sardà <pol...@gm...> del dia ds., 25
> de gen. 2025 a les 18:14:
>
>> Hi,
>>
>> in an application I develop, compilers tend to emit from time to time
>> SSE4a instructions as I target AMD CPUs. I filed a bug (486398) a while
>> ago, but given that I faced this same issue recently, I guessed that there
>> simply hadn't been bandwidth for a more than a decade old niche set of
>> instructions.
>>
>> Given that I still need them, and that there are only 4 instructions, I
>> have already added support for the INSERTQ and EXTRQ into VEX, and I intend
>> to finish the job by implementing MOVNTSD & MOVNTSS.
>>
>> Looking at the documentation, it seems to assume that you need to have
>> write access into the repo to publish code, so I'd like to request some
>> assistance regarding the logistics of submitting a patch (or, if possible,
>> creating a pull request).
>>
>> The documentation also mentions that I should use clang-format, but after
>> applying the formatting all the code changed, making a diff completely
>> unreadable. May I know if this is expected? Maybe VEX has been touched on
>> less, and hasn't had its formatting updated yet? Did I configure something
>> wrong?
>>
>> I'm certain that I'll have gotten some subtleties of Valgrind wrong, so
>> once I do submit a patch, please, review it thoroughly! :D
>>
>> Regards,
>>
>> Pol Marcet Sardà
>>
>>

Re: [Valgrind-developers] Request to contribute on SSE4a support

[Pol M. S. <pol...@gm...>]

Hello,

I'd need some assistance on how to submit a patch to the sourceware repo,
at the moment I have a patch cleaned up with all instructions implemented.
I am unfamiliar with the process (I am used to the PR mechanisms of
bitbucket/gitlab/github)

Sorry for writing a similar message to both the mailing list and the IRC

Best regards,

Pol Marcet Sardà

Missatge de Pol Marcet Sardà <pol...@gm...> del dia ds., 25 de
gen. 2025 a les 18:14:

> Hi,
>
> in an application I develop, compilers tend to emit from time to time
> SSE4a instructions as I target AMD CPUs. I filed a bug (486398) a while
> ago, but given that I faced this same issue recently, I guessed that there
> simply hadn't been bandwidth for a more than a decade old niche set of
> instructions.
>
> Given that I still need them, and that there are only 4 instructions, I
> have already added support for the INSERTQ and EXTRQ into VEX, and I intend
> to finish the job by implementing MOVNTSD & MOVNTSS.
>
> Looking at the documentation, it seems to assume that you need to have
> write access into the repo to publish code, so I'd like to request some
> assistance regarding the logistics of submitting a patch (or, if possible,
> creating a pull request).
>
> The documentation also mentions that I should use clang-format, but after
> applying the formatting all the code changed, making a diff completely
> unreadable. May I know if this is expected? Maybe VEX has been touched on
> less, and hasn't had its formatting updated yet? Did I configure something
> wrong?
>
> I'm certain that I'll have gotten some subtleties of Valgrind wrong, so
> once I do submit a patch, please, review it thoroughly! :D
>
> Regards,
>
> Pol Marcet Sardà
>
>