[PATCH 0/4] Many fixes to gdb.reverse tests

[PATCH 0/4] Many fixes to gdb.reverse tests

[Bruno Larsen]

This patch series applies multiple simple fixes to tests in the
gdb.reverse subdirectory. Changes were tested with the base
configuration in x86_64, both with clang and gcc, and no regressions
were found.

With these changes, there are only 10 unexpected failures extra when
running the subdirectory with clang, all from finish-reverse-next.exp,
and all relating to the issues that Carl Love has been trying to solve
with reverse stepping in contiguous ranges.

Bruno Larsen (4):
  gdb/testsuite: Fix many errors in gdb.reverse with clang
  gdb/testsuite: fix gdb.reverse/solib-*.exp tests when using clang
  gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang
  gdb/testsuite: Multiple improvements for gdb.reverse/insn-reverse.exp

 gdb/testsuite/gdb.reverse/break-reverse.c    |   4 +-
 gdb/testsuite/gdb.reverse/finish-reverse.c   |   4 +-
 gdb/testsuite/gdb.reverse/insn-reverse-x86.c | 324 +++++++++----------
 gdb/testsuite/gdb.reverse/insn-reverse.exp   |   4 +-
 gdb/testsuite/gdb.reverse/shr2.c             |   3 +-
 gdb/testsuite/gdb.reverse/solib-precsave.exp |  24 +-
 gdb/testsuite/gdb.reverse/solib-reverse.c    |   5 +-
 gdb/testsuite/gdb.reverse/solib-reverse.exp  |  24 +-
 gdb/testsuite/gdb.reverse/step-precsave.exp  |  18 +-
 gdb/testsuite/gdb.reverse/step-reverse.exp   |  20 +-
 gdb/testsuite/gdb.reverse/watch-reverse.c    |   4 +-
 gdb/testsuite/lib/gdb.exp                    |   6 +-
 12 files changed, 255 insertions(+), 185 deletions(-)

-- 
2.41.0

[PATCH 1/4] gdb/testsuite: Fix many errors in gdb.reverse with clang

[Bruno Larsen]

Clang does not add line information for lines that only contain a
closing } in functions. Many tests in the gdb.reverse folder set a
breakpoint in that line, but don't seem to use information available
after the return statement is executed, so this commit moves the
breakpoint to the previous line, where the return statement is.
---
 gdb/testsuite/gdb.reverse/break-reverse.c  | 4 ++--
 gdb/testsuite/gdb.reverse/finish-reverse.c | 4 ++--
 gdb/testsuite/gdb.reverse/watch-reverse.c  | 4 ++--
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/break-reverse.c b/gdb/testsuite/gdb.reverse/break-reverse.c
index 5f54a8f2bd7..4ae459f1bba 100644
--- a/gdb/testsuite/gdb.reverse/break-reverse.c
+++ b/gdb/testsuite/gdb.reverse/break-reverse.c
@@ -33,5 +33,5 @@ int main ()
 {
   xyz = 0;	/* break in main */
   foo ();
-  return (xyz == 2 ? 0 : 1);
-}		/* end of main */
+  return (xyz == 2 ? 0 : 1);	/* end of main */
+}
diff --git a/gdb/testsuite/gdb.reverse/finish-reverse.c b/gdb/testsuite/gdb.reverse/finish-reverse.c
index 609a14568a5..cc056ecfa8a 100644
--- a/gdb/testsuite/gdb.reverse/finish-reverse.c
+++ b/gdb/testsuite/gdb.reverse/finish-reverse.c
@@ -123,6 +123,6 @@ int main (int argc, char **argv)
   testval.double_testval = 3.14159265358979323846; /* float_checkpoint */
   double_resultval    = double_func ();		
   main_test = 1;				/* double_checkpoint */
-  return 0;
-} /* end of main */
+  return 0;					/* end of main */
+}
 
diff --git a/gdb/testsuite/gdb.reverse/watch-reverse.c b/gdb/testsuite/gdb.reverse/watch-reverse.c
index 8b7d1be0db5..e1b024af10b 100644
--- a/gdb/testsuite/gdb.reverse/watch-reverse.c
+++ b/gdb/testsuite/gdb.reverse/watch-reverse.c
@@ -206,6 +206,6 @@ int main ()
 
   func4 ();
 
-  return 0;
-} /* end of main */
+  return 0; /* end of main */
+}
 
-- 
2.41.0

[PATCH 2/4] gdb/testsuite: fix gdb.reverse/solib-*.exp tests when using clang

[Bruno Larsen]

The tests gdb.reverse/solib-precsave.exp and solib-reverse.exp have the
assumption that line tables will have an entry for the closing } in a
function. Not all compiles do this, one example being clang. To fix
this, this commit changes the function in shr2.c to have multiple lines,
and the test to accept either line as a correct step location.

To properly re-sync the inferiors, the function repeat_cmd_until had to
be slightly changed to work with empty "current locations", so that we
are able to step through multiple lines.

This also changes the annotations used to determine the breakpoint
locations in solib-reverse.c, adding a NOP right before the return
statement. This way GDB will not set a breakpoint in the closing } line.
---
 gdb/testsuite/gdb.reverse/shr2.c             |  3 ++-
 gdb/testsuite/gdb.reverse/solib-precsave.exp | 24 ++++++++++++++++----
 gdb/testsuite/gdb.reverse/solib-reverse.c    |  5 ++--
 gdb/testsuite/gdb.reverse/solib-reverse.exp  | 24 ++++++++++++++++----
 gdb/testsuite/lib/gdb.exp                    |  6 ++---
 5 files changed, 48 insertions(+), 14 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/shr2.c b/gdb/testsuite/gdb.reverse/shr2.c
index 84a03de1b8c..16c3bf3bc6c 100644
--- a/gdb/testsuite/gdb.reverse/shr2.c
+++ b/gdb/testsuite/gdb.reverse/shr2.c
@@ -19,7 +19,8 @@
 
 int shr2(int x)
 {
-  return 2*x;
+  int y = 2*x;
+  return y;
 }
 
 int shr2_local(int x)
diff --git a/gdb/testsuite/gdb.reverse/solib-precsave.exp b/gdb/testsuite/gdb.reverse/solib-precsave.exp
index 3ca73828063..f89969222c6 100644
--- a/gdb/testsuite/gdb.reverse/solib-precsave.exp
+++ b/gdb/testsuite/gdb.reverse/solib-precsave.exp
@@ -140,8 +140,17 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function one" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function one"
-gdb_test "reverse-step" " middle part two.*" "reverse-step back to main one"
+# Depending on wether the closing } has a line associated, we might have
+# different acceptable results here
+gdb_test_multiple "reverse-step" "reverse-step within solib function one" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main one"
 
 gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
     -re -wrap "begin part two.*" {
@@ -152,8 +161,15 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function two"
-gdb_test "reverse-step" " begin part two.*" "reverse-step back to main two"
+gdb_test_multiple "reverse-step" "reverse-step within solib function two" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main two"
 
 #
 # Test reverse-next over debuggable solib function
diff --git a/gdb/testsuite/gdb.reverse/solib-reverse.c b/gdb/testsuite/gdb.reverse/solib-reverse.c
index 7aa60089df3..f8c98f2ab94 100644
--- a/gdb/testsuite/gdb.reverse/solib-reverse.c
+++ b/gdb/testsuite/gdb.reverse/solib-reverse.c
@@ -43,6 +43,7 @@ int main ()
   shr1 ("message 2\n");		/* shr1 two */
   shr1 ("message 3\n");		/* shr1 three */
 
-  return 0;			/* end part one */
-} /* end of main */
+  asm("nop");			/* end part one */
+  return 0; /* end of main */
+}
 
diff --git a/gdb/testsuite/gdb.reverse/solib-reverse.exp b/gdb/testsuite/gdb.reverse/solib-reverse.exp
index c061086a8d7..afc4659184a 100644
--- a/gdb/testsuite/gdb.reverse/solib-reverse.exp
+++ b/gdb/testsuite/gdb.reverse/solib-reverse.exp
@@ -116,8 +116,17 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function one" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function one"
-gdb_test "reverse-step" " middle part two.*" "reverse-step back to main one"
+# Depending on wether the closing } has a line associated, we might have
+# different acceptable results here
+gdb_test_multiple "reverse-step" "reverse-step within solib function one" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main one"
 
 gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
     -re -wrap "begin part two.*" {
@@ -128,8 +137,15 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function two"
-gdb_test "reverse-step" " begin part two.*" "reverse-step back to main two"
+gdb_test_multiple "reverse-step" "reverse-step within solib function two" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main two"
 
 #
 # Test reverse-next over debuggable solib function
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 66c04a2efe2..63b6291fc36 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -9679,6 +9679,9 @@ proc repeat_cmd_until { command current target \
 
     set count 0
     gdb_test_multiple "$command" "$test_name" {
+	-re "$target.*$gdb_prompt $" {
+	    pass "$test_name"
+	}
 	-re "$current.*$gdb_prompt $" {
 	    incr count
 	    if { $count < $max_steps } {
@@ -9688,9 +9691,6 @@ proc repeat_cmd_until { command current target \
 		fail "$test_name"
 	    }
 	}
-	-re "$target.*$gdb_prompt $" {
-	    pass "$test_name"
-	}
     }
 }
 
-- 
2.41.0

[PATCH 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Bruno Larsen]

When testing using reverse-stepi to fully step through a function, the
code checks for an infinite loop by seeing if we land on the line that
contains the return statement multiple times. This assumption only works
if there is only one instruction associated with that line, which is how
GCC handles line information, but other compilers may handle it differently.
Clang-15, for instance, associates 6. Because of this, the inferior used
to get seriously out of sync with the test expectations, and result in 13
spurious failures. The same issue occurs with gdb.reverse/step-precsave.exp.

This commit changes the test so that we check for PC instead of line
number. The test still only happens when the same line is detected, to
simplify the resulting log. With this change, no new failures are
emitted when using clang.
---
 gdb/testsuite/gdb.reverse/step-precsave.exp | 18 +++++++++++++++++-
 gdb/testsuite/gdb.reverse/step-reverse.exp  | 20 +++++++++++++++++++-
 2 files changed, 36 insertions(+), 2 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/step-precsave.exp b/gdb/testsuite/gdb.reverse/step-precsave.exp
index 19cd5d9930e..da3a47e07e2 100644
--- a/gdb/testsuite/gdb.reverse/step-precsave.exp
+++ b/gdb/testsuite/gdb.reverse/step-precsave.exp
@@ -30,6 +30,16 @@ if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
     return -1
 }
 
+proc get_current_pc {} {
+    set pc 0
+    gdb_test_multiple "print \$pc" "" {
+	-re -wrap ".*0x(\[0-9a-f\]+).*" {
+	    set pc $expect_out(1,string)
+	}
+    }
+    return $pc
+}
+
 runto_main
 
 # Activate process record/replay
@@ -209,11 +219,17 @@ gdb_test_multiple "stepi" "$test_message" {
 
 # stepi backward out of a function call
 
+set start_pc [get_current_pc]
 set stepi_location  [gdb_get_line_number "STEPI TEST" "$srcfile"]
 set test_message "reverse stepi from a function call"
 gdb_test_multiple "stepi" "$test_message" {
     -re "ARRIVED IN CALLEE.*$gdb_prompt $" {
-	fail "$test_message (start statement)"
+	if { [get_current_pc] == $start_pc } {
+	    fail "$test_message (start statement)"
+	} else {
+	    send_gdb "stepi\n" 
+	    exp_continue
+	}
     }
     -re "ENTER CALLEE.*$gdb_prompt $" {
 	send_gdb "stepi\n" 
diff --git a/gdb/testsuite/gdb.reverse/step-reverse.exp b/gdb/testsuite/gdb.reverse/step-reverse.exp
index 4b78a8f8fb7..1c45c6f645a 100644
--- a/gdb/testsuite/gdb.reverse/step-reverse.exp
+++ b/gdb/testsuite/gdb.reverse/step-reverse.exp
@@ -28,6 +28,16 @@ if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
     return -1
 }
 
+proc get_current_pc {} {
+    set pc 0
+    gdb_test_multiple "print \$pc" "" {
+	-re -wrap ".*0x(\[0-9a-f\]+).*" {
+	    set pc $expect_out(1,string)
+	}
+    }
+    return $pc
+}
+
 runto_main
 
 if [supports_process_record] {
@@ -174,11 +184,19 @@ gdb_test_multiple "stepi" "$test_message" {
 
 # stepi backward out of a function call
 
+# When testing stepi, we dont want to infinitely step if we're not moving
+# so we store the starting PC, in case we land on the same line as above
+set start_pc [get_current_pc]
 set stepi_location  [gdb_get_line_number "STEPI TEST" "$srcfile"]
 set test_message "reverse stepi from a function call"
 gdb_test_multiple "stepi" "$test_message" {
     -re "ARRIVED IN CALLEE.*$gdb_prompt $" {
-	fail "$test_message (start statement)"
+	if { [get_current_pc] == $start_pc } {
+	    fail "$test_message (start statement)"
+	} else {
+	    send_gdb "stepi\n" 
+	    exp_continue
+	}
     }
     -re "ENTER CALLEE.*$gdb_prompt $" {
 	send_gdb "stepi\n" 
-- 
2.41.0

[PATCH 4/4] gdb/testsuite: Multiple improvements for gdb.reverse/insn-reverse.exp

[Bruno Larsen]

This commits tackles 2 problems in the test
gdb.reverse/insn-reverse.exp. They are, broadly: flawed logic when an
unexpected error occurs, and badly formed asm expressions.

For the first, what happens is that if the inferior stops progressing
for some reason, the test will emit an UNSUPPORTED and continue testing
by reversing from the current location and checking all registers for
every instruction.  However, due to how the outputs are indexed in the
test, this early exit will cause most of the subsequent tests to be
de-synced and will emit many unrelated failures.

This commit changes the UNSUPPORTED for a FAIL, since the test has in
fact failed to record the execution of the whole function, and
decrements the recorded instruction count by one so that the indexes are
in sync once more.

At the time of committing, this reduces the amount of failures when
testing with clang-15 from around 150 to 2, and correctly identifies
where the issue lies.

The second problem is in how the asm statements in the *-x86.c file
are written. As an example, let's examine the following line:

__asm__ volatile ("rdrand %%ebp;" : "=r" (number));

This statement says that number is being used as the output variable,
but is not indicating which registers were clobbered so that the
compiler is able to properly output. GCC decides to just not save
anything, whereas clang assumes that the output is in %rax, and writes
it to the variable. This hid the problem that any compiler is not good
at dealing with asm statements that change the rbp register. It can be
seen more explicitly by informing gcc that rbp has been clobbered like
so:

__asm__ volatile ("rdrand %%ebp;" : "=r" (number) : : "%ebp");

This statement gets compiled into the following assembly:
        rdrandl %ebp
        movl    %eax, -4(%rbp)

Which is clearly using the incorrect rbp to find the memory location of
the variable. Since the test only exercises GDB's ability to record the
register changes, this commit removes the output to memory.

Finally, correctly informing the compiler of clobbered registers
makes gcc throw an error that the rsp is no longer usable at the end of
the function. To avoid that, this commit compresses the 3 asm statements
that would save, change and reset registers into a single asm statement.
---
 gdb/testsuite/gdb.reverse/insn-reverse-x86.c | 324 +++++++++----------
 gdb/testsuite/gdb.reverse/insn-reverse.exp   |   4 +-
 2 files changed, 165 insertions(+), 163 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/insn-reverse-x86.c b/gdb/testsuite/gdb.reverse/insn-reverse-x86.c
index da9999e9942..2b4fb4c10e0 100644
--- a/gdb/testsuite/gdb.reverse/insn-reverse-x86.c
+++ b/gdb/testsuite/gdb.reverse/insn-reverse-x86.c
@@ -64,100 +64,100 @@ rdrand (void)
     return;
 
   /* 16-bit random numbers.  */
-  __asm__ volatile ("rdrand %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%bx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%cx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%dx;" : "=r" (number));
+  __asm__ volatile ("rdrand %%ax;": : : "%ax");
+  __asm__ volatile ("rdrand %%bx;": : : "%bx");
+  __asm__ volatile ("rdrand %%cx;": : : "%cx");
+  __asm__ volatile ("rdrand %%dx;": : : "%dx");
 
-  __asm__ volatile ("mov %%di, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%di;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%di;" : "=r" (number));
+  __asm__ volatile ("mov %%di, %%ax;\n\
+		     rdrand %%di;\n\
+		     mov %%ax, %%di;" : : : "%ax");
 
-  __asm__ volatile ("mov %%si, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%si;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%si;" : "=r" (number));
+  __asm__ volatile ("mov %%si, %%ax;\n\
+		     rdrand %%si;\n\
+		     mov %%ax, %%si;" : : : "%ax");
 
-  __asm__ volatile ("mov %%bp, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%bp;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%bp;" : "=r" (number));
+  __asm__ volatile ("mov %%bp, %%ax;\n\
+		     rdrand %%bp;\n\
+		     mov %%ax, %%bp;" : : : "%ax");
 
-  __asm__ volatile ("mov %%sp, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%sp;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%sp;" : "=r" (number));
+  __asm__ volatile ("mov %%sp, %%ax;\n\
+		     rdrand %%sp;\n\
+		     mov %%ax, %%sp;" : : : "%ax");
 
 #ifdef __x86_64__
-  __asm__ volatile ("rdrand %%r8w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r9w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r10w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r11w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r12w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r13w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r14w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r15w;" : "=r" (number));
+  __asm__ volatile ("rdrand %%r8w;": : : "%r8");
+  __asm__ volatile ("rdrand %%r9w;": : : "%r9");
+  __asm__ volatile ("rdrand %%r10w;": : : "%r10");
+  __asm__ volatile ("rdrand %%r11w;": : : "%r11");
+  __asm__ volatile ("rdrand %%r12w;": : : "%r12");
+  __asm__ volatile ("rdrand %%r13w;": : : "%r13");
+  __asm__ volatile ("rdrand %%r14w;": : : "%r14");
+  __asm__ volatile ("rdrand %%r15w;": : : "%r15");
 #endif
 
   /* 32-bit random numbers.  */
-  __asm__ volatile ("rdrand %%eax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%ebx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%ecx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%edx;" : "=r" (number));
+  __asm__ volatile ("rdrand %%eax;": : : "%eax");
+  __asm__ volatile ("rdrand %%ebx;": : : "%ebx");
+  __asm__ volatile ("rdrand %%ecx;": : : "%ecx");
+  __asm__ volatile ("rdrand %%edx;": : : "%edx");
 
 #ifdef __x86_64__
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%edi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%esi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%ebp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%esp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (number));
-
-  __asm__ volatile ("rdrand %%r8d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r9d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r10d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r11d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r12d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r13d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r14d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r15d;" : "=r" (number));
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdrand %%edi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdrand %%esi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdrand %%ebp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdrand %%esp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdrand %%r8d;": : : "%r8");
+  __asm__ volatile ("rdrand %%r9d;": : : "%r9");
+  __asm__ volatile ("rdrand %%r10d;": : : "%r10");
+  __asm__ volatile ("rdrand %%r11d;": : : "%r11");
+  __asm__ volatile ("rdrand %%r12d;": : : "%r12");
+  __asm__ volatile ("rdrand %%r13d;": : : "%r13");
+  __asm__ volatile ("rdrand %%r14d;": : : "%r14");
+  __asm__ volatile ("rdrand %%r15d;": : : "%r15");
 
   /* 64-bit random numbers.  */
-  __asm__ volatile ("rdrand %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rbx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rcx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rdx;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rdi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rsi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rbp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rsp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (number));
-
-  __asm__ volatile ("rdrand %%r8;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r9;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r10;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r11;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r12;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r13;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r14;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r15;" : "=r" (number));
+  __asm__ volatile ("rdrand %%rax;": : : "%rax");
+  __asm__ volatile ("rdrand %%rbx;": : : "%rbx");
+  __asm__ volatile ("rdrand %%rcx;": : : "%rcx");
+  __asm__ volatile ("rdrand %%rdx;": : : "%rdx");
+
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdrand %%rdi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdrand %%rsi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdrand %%rbp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdrand %%rsp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdrand %%r8;": : : "%r8");
+  __asm__ volatile ("rdrand %%r9;": : : "%r9");
+  __asm__ volatile ("rdrand %%r10;": : : "%r10");
+  __asm__ volatile ("rdrand %%r11;": : : "%r11");
+  __asm__ volatile ("rdrand %%r12;": : : "%r12");
+  __asm__ volatile ("rdrand %%r13;": : : "%r13");
+  __asm__ volatile ("rdrand %%r14;": : : "%r14");
+  __asm__ volatile ("rdrand %%r15;": : : "%r15");
 #endif
 }
 
@@ -173,100 +173,100 @@ rdseed (void)
     return;
 
   /* 16-bit random seeds.  */
-  __asm__ volatile ("rdseed %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%bx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%cx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%dx;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%ax;": : : "%ax");
+  __asm__ volatile ("rdseed %%bx;": : : "%bx");
+  __asm__ volatile ("rdseed %%cx;": : : "%cx");
+  __asm__ volatile ("rdseed %%dx;": : : "%dx");
 
-  __asm__ volatile ("mov %%di, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%di;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%di;" : "=r" (seed));
+  __asm__ volatile ("mov %%di, %%ax;\n\
+		     rdseed %%di;\n\
+		     mov %%ax, %%di;" : : : "%ax");
 
-  __asm__ volatile ("mov %%si, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%si;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%si;" : "=r" (seed));
+  __asm__ volatile ("mov %%si, %%ax;\n\
+		     rdseed %%si;\n\
+		     mov %%ax, %%si;" : : : "%ax");
 
-  __asm__ volatile ("mov %%bp, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%bp;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%bp;" : "=r" (seed));
+  __asm__ volatile ("mov %%bp, %%ax;\n\
+		     rdseed %%bp;\n\
+		     mov %%ax, %%bp;" : : : "%ax");
 
-  __asm__ volatile ("mov %%sp, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%sp;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%sp;" : "=r" (seed));
+  __asm__ volatile ("mov %%sp, %%ax;\n\
+		     rdseed %%sp;\n\
+		     mov %%ax, %%sp;" : : : "%ax");
 
 #ifdef __x86_64__
-  __asm__ volatile ("rdseed %%r8w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r9w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r10w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r11w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r12w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r13w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r14w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r15w;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%r8w;": : : "%r8");
+  __asm__ volatile ("rdseed %%r9w;": : : "%r9");
+  __asm__ volatile ("rdseed %%r10w;": : : "%r10");
+  __asm__ volatile ("rdseed %%r11w;": : : "%r11");
+  __asm__ volatile ("rdseed %%r12w;": : : "%r12");
+  __asm__ volatile ("rdseed %%r13w;": : : "%r13");
+  __asm__ volatile ("rdseed %%r14w;": : : "%r14");
+  __asm__ volatile ("rdseed %%r15w;": : : "%r15");
 #endif
 
   /* 32-bit random seeds.  */
-  __asm__ volatile ("rdseed %%eax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%ebx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%ecx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%edx;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%eax;": : : "%eax");
+  __asm__ volatile ("rdseed %%ebx;": : : "%ebx");
+  __asm__ volatile ("rdseed %%ecx;": : : "%ecx");
+  __asm__ volatile ("rdseed %%edx;": : : "%edx");
 
 #ifdef __x86_64__
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%edi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%esi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%ebp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%esp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (seed));
-
-  __asm__ volatile ("rdseed %%r8d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r9d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r10d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r11d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r12d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r13d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r14d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r15d;" : "=r" (seed));
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdseed %%edi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdseed %%esi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdseed %%ebp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdseed %%esp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdseed %%r8d;": : : "%r8");
+  __asm__ volatile ("rdseed %%r9d;": : : "%r9");
+  __asm__ volatile ("rdseed %%r10d;": : : "%r10");
+  __asm__ volatile ("rdseed %%r11d;": : : "%r11");
+  __asm__ volatile ("rdseed %%r12d;": : : "%r12");
+  __asm__ volatile ("rdseed %%r13d;": : : "%r13");
+  __asm__ volatile ("rdseed %%r14d;": : : "%r14");
+  __asm__ volatile ("rdseed %%r15d;": : : "%r15");
 
   /* 64-bit random seeds.  */
-  __asm__ volatile ("rdseed %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rbx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rcx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rdx;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rdi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rsi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rbp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rsp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (seed));
-
-  __asm__ volatile ("rdseed %%r8;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r9;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r10;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r11;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r12;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r13;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r14;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r15;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%rax;": : : "%rax");
+  __asm__ volatile ("rdseed %%rbx;": : : "%rbx");
+  __asm__ volatile ("rdseed %%rcx;": : : "%rcx");
+  __asm__ volatile ("rdseed %%rdx;": : : "%rdx");
+
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdseed %%rdi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdseed %%rsi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdseed %%rbp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdseed %%rsp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdseed %%r8;": : : "%r8");
+  __asm__ volatile ("rdseed %%r9;": : : "%r9");
+  __asm__ volatile ("rdseed %%r10;": : : "%r10");
+  __asm__ volatile ("rdseed %%r11;": : : "%r11");
+  __asm__ volatile ("rdseed %%r12;": : : "%r12");
+  __asm__ volatile ("rdseed %%r13;": : : "%r13");
+  __asm__ volatile ("rdseed %%r14;": : : "%r14");
+  __asm__ volatile ("rdseed %%r15;": : : "%r15");
 #endif
 }
 
diff --git a/gdb/testsuite/gdb.reverse/insn-reverse.exp b/gdb/testsuite/gdb.reverse/insn-reverse.exp
index 1a575b2d43e..206e765dd50 100644
--- a/gdb/testsuite/gdb.reverse/insn-reverse.exp
+++ b/gdb/testsuite/gdb.reverse/insn-reverse.exp
@@ -95,7 +95,9 @@ proc test { func testcase_nr } {
 
 	    if { $prev_insn_addr == $insn_addr } {
 		# Failed to make progress, might have run into SIGILL.
-		unsupported "no progress at: $expect_out(2,string)"
+		fail "no progress at: $expect_out(2,string)"
+		# Ignore the last instruction recorded
+		incr count -1
 		break
 	    }
 
-- 
2.41.0

Re: [PATCH 2/4] gdb/testsuite: fix gdb.reverse/solib-*.exp tests when using clang

[Tom Tromey]

>>>>> "Bruno" == Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:

Bruno> diff --git a/gdb/testsuite/gdb.reverse/solib-reverse.c b/gdb/testsuite/gdb.reverse/solib-reverse.c
Bruno> index 7aa60089df3..f8c98f2ab94 100644
Bruno> --- a/gdb/testsuite/gdb.reverse/solib-reverse.c
Bruno> +++ b/gdb/testsuite/gdb.reverse/solib-reverse.c
Bruno> @@ -43,6 +43,7 @@ int main ()
Bruno>    shr1 ("message 2\n");		/* shr1 two */
Bruno>    shr1 ("message 3\n");		/* shr1 three */
 
Bruno> -  return 0;			/* end part one */
Bruno> -} /* end of main */
Bruno> +  asm("nop");			/* end part one */
Bruno> +  return 0; /* end of main */

I suspect it would be better to use something other than 'asm' here,
like an assignment to some global or something like that.

Tom

Re: [PATCH 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Tom Tromey]

>>>>> "Bruno" == Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:

Bruno> +proc get_current_pc {} {

I think this can probably use get_hexadecimal_valueof instead.

Tom

[PATCH v2 0/4] Many fixes to gdb.reverse tests

[Bruno Larsen]

This patch series applies multiple simple fixes to tests in the
gdb.reverse subdirectory. Changes were tested with the base
configuration in x86_64, both with clang and gcc, and no regressions
were found.

With these changes, there are only 10 unexpected failures extra when
running the subdirectory with clang, all from finish-reverse-next.exp,
and all relating to the issues that Carl Love has been trying to solve
with reverse stepping in contiguous ranges.

Changes for v2:
 * Implemented Tromey's suggestions
 * Small change to get_hexadecimal_valueof - added nopass option

Bruno Larsen (4):
  gdb/testsuite: Fix many errors in gdb.reverse with clang
  gdb/testsuite: fix gdb.reverse/solib-*.exp tests when using clang
  gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang
  gdb/testsuite: Multiple improvements for gdb.reverse/insn-reverse.exp

 gdb/testsuite/gdb.reverse/break-reverse.c    |   4 +-
 gdb/testsuite/gdb.reverse/finish-reverse.c   |   4 +-
 gdb/testsuite/gdb.reverse/insn-reverse-x86.c | 324 +++++++++----------
 gdb/testsuite/gdb.reverse/insn-reverse.exp   |   4 +-
 gdb/testsuite/gdb.reverse/shr2.c             |   3 +-
 gdb/testsuite/gdb.reverse/solib-precsave.exp |  24 +-
 gdb/testsuite/gdb.reverse/solib-reverse.c    |   5 +-
 gdb/testsuite/gdb.reverse/solib-reverse.exp  |  24 +-
 gdb/testsuite/gdb.reverse/step-precsave.exp  |  18 +-
 gdb/testsuite/gdb.reverse/step-reverse.exp   |  20 +-
 gdb/testsuite/gdb.reverse/watch-reverse.c    |   4 +-
 gdb/testsuite/lib/gdb.exp                    |  12 +-
 12 files changed, 259 insertions(+), 187 deletions(-)

-- 
2.41.0

[PATCH v2 1/4] gdb/testsuite: Fix many errors in gdb.reverse with clang

[Bruno Larsen]

Clang does not add line information for lines that only contain a
closing } in functions. Many tests in the gdb.reverse folder set a
breakpoint in that line, but don't seem to use information available
after the return statement is executed, so this commit moves the
breakpoint to the previous line, where the return statement is.
---
 gdb/testsuite/gdb.reverse/break-reverse.c  | 4 ++--
 gdb/testsuite/gdb.reverse/finish-reverse.c | 4 ++--
 gdb/testsuite/gdb.reverse/watch-reverse.c  | 4 ++--
 3 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/break-reverse.c b/gdb/testsuite/gdb.reverse/break-reverse.c
index 5f54a8f2bd7..4ae459f1bba 100644
--- a/gdb/testsuite/gdb.reverse/break-reverse.c
+++ b/gdb/testsuite/gdb.reverse/break-reverse.c
@@ -33,5 +33,5 @@ int main ()
 {
   xyz = 0;	/* break in main */
   foo ();
-  return (xyz == 2 ? 0 : 1);
-}		/* end of main */
+  return (xyz == 2 ? 0 : 1);	/* end of main */
+}
diff --git a/gdb/testsuite/gdb.reverse/finish-reverse.c b/gdb/testsuite/gdb.reverse/finish-reverse.c
index 609a14568a5..cc056ecfa8a 100644
--- a/gdb/testsuite/gdb.reverse/finish-reverse.c
+++ b/gdb/testsuite/gdb.reverse/finish-reverse.c
@@ -123,6 +123,6 @@ int main (int argc, char **argv)
   testval.double_testval = 3.14159265358979323846; /* float_checkpoint */
   double_resultval    = double_func ();		
   main_test = 1;				/* double_checkpoint */
-  return 0;
-} /* end of main */
+  return 0;					/* end of main */
+}
 
diff --git a/gdb/testsuite/gdb.reverse/watch-reverse.c b/gdb/testsuite/gdb.reverse/watch-reverse.c
index 8b7d1be0db5..e1b024af10b 100644
--- a/gdb/testsuite/gdb.reverse/watch-reverse.c
+++ b/gdb/testsuite/gdb.reverse/watch-reverse.c
@@ -206,6 +206,6 @@ int main ()
 
   func4 ();
 
-  return 0;
-} /* end of main */
+  return 0; /* end of main */
+}
 
-- 
2.41.0

[PATCH v2 2/4] gdb/testsuite: fix gdb.reverse/solib-*.exp tests when using clang

[Bruno Larsen]

The tests gdb.reverse/solib-precsave.exp and solib-reverse.exp have the
assumption that line tables will have an entry for the closing } in a
function. Not all compiles do this, one example being clang. To fix
this, this commit changes the function in shr2.c to have multiple lines,
and the test to accept either line as a correct step location.

To properly re-sync the inferiors, the function repeat_cmd_until had to
be slightly changed to work with empty "current locations", so that we
are able to step through multiple lines.

This also changes the annotations used to determine the breakpoint
locations in solib-reverse.c, adding a simple variable assignment right
before the return statement. This way GDB will not set a breakpoint in
the closing } line.
---
 gdb/testsuite/gdb.reverse/shr2.c             |  3 ++-
 gdb/testsuite/gdb.reverse/solib-precsave.exp | 24 ++++++++++++++++----
 gdb/testsuite/gdb.reverse/solib-reverse.c    |  5 ++--
 gdb/testsuite/gdb.reverse/solib-reverse.exp  | 24 ++++++++++++++++----
 gdb/testsuite/lib/gdb.exp                    |  6 ++---
 5 files changed, 48 insertions(+), 14 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/shr2.c b/gdb/testsuite/gdb.reverse/shr2.c
index 84a03de1b8c..16c3bf3bc6c 100644
--- a/gdb/testsuite/gdb.reverse/shr2.c
+++ b/gdb/testsuite/gdb.reverse/shr2.c
@@ -19,7 +19,8 @@
 
 int shr2(int x)
 {
-  return 2*x;
+  int y = 2*x;
+  return y;
 }
 
 int shr2_local(int x)
diff --git a/gdb/testsuite/gdb.reverse/solib-precsave.exp b/gdb/testsuite/gdb.reverse/solib-precsave.exp
index 3ca73828063..f89969222c6 100644
--- a/gdb/testsuite/gdb.reverse/solib-precsave.exp
+++ b/gdb/testsuite/gdb.reverse/solib-precsave.exp
@@ -140,8 +140,17 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function one" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function one"
-gdb_test "reverse-step" " middle part two.*" "reverse-step back to main one"
+# Depending on wether the closing } has a line associated, we might have
+# different acceptable results here
+gdb_test_multiple "reverse-step" "reverse-step within solib function one" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main one"
 
 gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
     -re -wrap "begin part two.*" {
@@ -152,8 +161,15 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function two"
-gdb_test "reverse-step" " begin part two.*" "reverse-step back to main two"
+gdb_test_multiple "reverse-step" "reverse-step within solib function two" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main two"
 
 #
 # Test reverse-next over debuggable solib function
diff --git a/gdb/testsuite/gdb.reverse/solib-reverse.c b/gdb/testsuite/gdb.reverse/solib-reverse.c
index 7aa60089df3..20288e0bee8 100644
--- a/gdb/testsuite/gdb.reverse/solib-reverse.c
+++ b/gdb/testsuite/gdb.reverse/solib-reverse.c
@@ -43,6 +43,7 @@ int main ()
   shr1 ("message 2\n");		/* shr1 two */
   shr1 ("message 3\n");		/* shr1 three */
 
-  return 0;			/* end part one */
-} /* end of main */
+  b[0] = 0;			/* end part one */
+  return 0; /* end of main */
+}
 
diff --git a/gdb/testsuite/gdb.reverse/solib-reverse.exp b/gdb/testsuite/gdb.reverse/solib-reverse.exp
index c061086a8d7..afc4659184a 100644
--- a/gdb/testsuite/gdb.reverse/solib-reverse.exp
+++ b/gdb/testsuite/gdb.reverse/solib-reverse.exp
@@ -116,8 +116,17 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function one" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function one"
-gdb_test "reverse-step" " middle part two.*" "reverse-step back to main one"
+# Depending on wether the closing } has a line associated, we might have
+# different acceptable results here
+gdb_test_multiple "reverse-step" "reverse-step within solib function one" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main one"
 
 gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
     -re -wrap "begin part two.*" {
@@ -128,8 +137,15 @@ gdb_test_multiple "reverse-step" "reverse-step into solib function two" {
 	pass $gdb_test_name
     }
 }
-gdb_test "reverse-step" "return 2.x.*" "reverse-step within solib function two"
-gdb_test "reverse-step" " begin part two.*" "reverse-step back to main two"
+gdb_test_multiple "reverse-step" "reverse-step within solib function two" {
+    -re -wrap "return y;.*" {
+	pass $gdb_test_name
+    }
+    -re -wrap "int y =.*" {
+	pass $gdb_test_name
+    }
+}
+repeat_cmd_until "reverse-step" "" "main .. at" "reverse-step back to main two"
 
 #
 # Test reverse-next over debuggable solib function
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 66c04a2efe2..63b6291fc36 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -9679,6 +9679,9 @@ proc repeat_cmd_until { command current target \
 
     set count 0
     gdb_test_multiple "$command" "$test_name" {
+	-re "$target.*$gdb_prompt $" {
+	    pass "$test_name"
+	}
 	-re "$current.*$gdb_prompt $" {
 	    incr count
 	    if { $count < $max_steps } {
@@ -9688,9 +9691,6 @@ proc repeat_cmd_until { command current target \
 		fail "$test_name"
 	    }
 	}
-	-re "$target.*$gdb_prompt $" {
-	    pass "$test_name"
-	}
     }
 }
 
-- 
2.41.0

[PATCH v2 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Bruno Larsen]

When testing using reverse-stepi to fully step through a function, the
code checks for an infinite loop by seeing if we land on the line that
contains the return statement multiple times. This assumption only works
if there is only one instruction associated with that line, which is how
GCC handles line information, but other compilers may handle it differently.
Clang-15, for instance, associates 6. Because of this, the inferior used
to get seriously out of sync with the test expectations, and result in 13
spurious failures. The same issue occurs with gdb.reverse/step-precsave.exp.

This commit changes the test so that we check for PC instead of line
number. The test still only happens when the same line is detected, to
simplify the resulting log. With this change, no new failures are
emitted when using clang.

It also adds a new parameter to get_hexadecimal_valueof, so that we can
use it without generating new passes, otherwise we'd get multiple
duplicate test names. This change shouldn't affect any other test using
this proc.
---
 gdb/testsuite/gdb.reverse/step-precsave.exp | 18 +++++++++++++++++-
 gdb/testsuite/gdb.reverse/step-reverse.exp  | 20 +++++++++++++++++++-
 gdb/testsuite/lib/gdb.exp                   |  6 ++++--
 3 files changed, 40 insertions(+), 4 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/step-precsave.exp b/gdb/testsuite/gdb.reverse/step-precsave.exp
index 19cd5d9930e..da3a47e07e2 100644
--- a/gdb/testsuite/gdb.reverse/step-precsave.exp
+++ b/gdb/testsuite/gdb.reverse/step-precsave.exp
@@ -30,6 +30,16 @@ if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
     return -1
 }
 
+proc get_current_pc {} {
+    set pc 0
+    gdb_test_multiple "print \$pc" "" {
+	-re -wrap ".*0x(\[0-9a-f\]+).*" {
+	    set pc $expect_out(1,string)
+	}
+    }
+    return $pc
+}
+
 runto_main
 
 # Activate process record/replay
@@ -209,11 +219,17 @@ gdb_test_multiple "stepi" "$test_message" {
 
 # stepi backward out of a function call
 
+set start_pc [get_current_pc]
 set stepi_location  [gdb_get_line_number "STEPI TEST" "$srcfile"]
 set test_message "reverse stepi from a function call"
 gdb_test_multiple "stepi" "$test_message" {
     -re "ARRIVED IN CALLEE.*$gdb_prompt $" {
-	fail "$test_message (start statement)"
+	if { [get_current_pc] == $start_pc } {
+	    fail "$test_message (start statement)"
+	} else {
+	    send_gdb "stepi\n" 
+	    exp_continue
+	}
     }
     -re "ENTER CALLEE.*$gdb_prompt $" {
 	send_gdb "stepi\n" 
diff --git a/gdb/testsuite/gdb.reverse/step-reverse.exp b/gdb/testsuite/gdb.reverse/step-reverse.exp
index 4b78a8f8fb7..9ff97bfde42 100644
--- a/gdb/testsuite/gdb.reverse/step-reverse.exp
+++ b/gdb/testsuite/gdb.reverse/step-reverse.exp
@@ -28,6 +28,16 @@ if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
     return -1
 }
 
+proc get_current_pc {} {
+    set pc 0
+    gdb_test_multiple "print \$pc" "" {
+	-re -wrap ".*0x(\[0-9a-f\]+).*" {
+	    set pc $expect_out(1,string)
+	}
+    }
+    return $pc
+}
+
 runto_main
 
 if [supports_process_record] {
@@ -174,11 +184,19 @@ gdb_test_multiple "stepi" "$test_message" {
 
 # stepi backward out of a function call
 
+# When testing stepi, we dont want to infinitely step if we're not moving
+# so we store the starting PC, in case we land on the same line as above
+set start_pc [get_hexadecimal_valueof "\$pc" 0 "" "false"]
 set stepi_location  [gdb_get_line_number "STEPI TEST" "$srcfile"]
 set test_message "reverse stepi from a function call"
 gdb_test_multiple "stepi" "$test_message" {
     -re "ARRIVED IN CALLEE.*$gdb_prompt $" {
-	fail "$test_message (start statement)"
+	if { [get_hexadecimal_valueof "\$pc" 0 "" "false"] == $start_pc } {
+	    fail "$test_message (start statement)"
+	} else {
+	    send_gdb "stepi\n" 
+	    exp_continue
+	}
     }
     -re "ENTER CALLEE.*$gdb_prompt $" {
 	send_gdb "stepi\n" 
diff --git a/gdb/testsuite/lib/gdb.exp b/gdb/testsuite/lib/gdb.exp
index 63b6291fc36..37342583e0a 100644
--- a/gdb/testsuite/lib/gdb.exp
+++ b/gdb/testsuite/lib/gdb.exp
@@ -7919,7 +7919,7 @@ proc get_integer_valueof { exp default {test ""} } {
 # TEST is the test message to use.  It can be omitted, in which case
 # a test message is built from EXP.
 
-proc get_hexadecimal_valueof { exp default {test ""} } {
+proc get_hexadecimal_valueof { exp default {test ""} {pass true} } {
     global gdb_prompt
 
     if {$test == ""} {
@@ -7930,7 +7930,9 @@ proc get_hexadecimal_valueof { exp default {test ""} } {
     gdb_test_multiple "print /x ${exp}" $test {
 	-re "\\$\[0-9\]* = (0x\[0-9a-zA-Z\]+).*$gdb_prompt $" {
 	    set val $expect_out(1,string)
-	    pass "$test"
+	    if { "$pass" == "true" } {
+		pass "$test"
+	    }
 	}
     }
     return ${val}
-- 
2.41.0

[PATCH v2 4/4] gdb/testsuite: Multiple improvements for gdb.reverse/insn-reverse.exp

[Bruno Larsen]

This commits tackles 2 problems in the test
gdb.reverse/insn-reverse.exp. They are, broadly: flawed logic when an
unexpected error occurs, and badly formed asm expressions.

For the first, what happens is that if the inferior stops progressing
for some reason, the test will emit an UNSUPPORTED and continue testing
by reversing from the current location and checking all registers for
every instruction.  However, due to how the outputs are indexed in the
test, this early exit will cause most of the subsequent tests to be
de-synced and will emit many unrelated failures.

This commit changes the UNSUPPORTED for a FAIL, since the test has in
fact failed to record the execution of the whole function, and
decrements the recorded instruction count by one so that the indexes are
in sync once more.

At the time of committing, this reduces the amount of failures when
testing with clang-15 from around 150 to 2, and correctly identifies
where the issue lies.

The second problem is in how the asm statements in the *-x86.c file
are written. As an example, let's examine the following line:

__asm__ volatile ("rdrand %%ebp;" : "=r" (number));

This statement says that number is being used as the output variable,
but is not indicating which registers were clobbered so that the
compiler is able to properly output. GCC decides to just not save
anything, whereas clang assumes that the output is in %rax, and writes
it to the variable. This hid the problem that any compiler is not good
at dealing with asm statements that change the rbp register. It can be
seen more explicitly by informing gcc that rbp has been clobbered like
so:

__asm__ volatile ("rdrand %%ebp;" : "=r" (number) : : "%ebp");

This statement gets compiled into the following assembly:
        rdrandl %ebp
        movl    %eax, -4(%rbp)

Which is clearly using the incorrect rbp to find the memory location of
the variable. Since the test only exercises GDB's ability to record the
register changes, this commit removes the output to memory.

Finally, correctly informing the compiler of clobbered registers
makes gcc throw an error that the rsp is no longer usable at the end of
the function. To avoid that, this commit compresses the 3 asm statements
that would save, change and reset registers into a single asm statement.
---
 gdb/testsuite/gdb.reverse/insn-reverse-x86.c | 324 +++++++++----------
 gdb/testsuite/gdb.reverse/insn-reverse.exp   |   4 +-
 2 files changed, 165 insertions(+), 163 deletions(-)

diff --git a/gdb/testsuite/gdb.reverse/insn-reverse-x86.c b/gdb/testsuite/gdb.reverse/insn-reverse-x86.c
index da9999e9942..2b4fb4c10e0 100644
--- a/gdb/testsuite/gdb.reverse/insn-reverse-x86.c
+++ b/gdb/testsuite/gdb.reverse/insn-reverse-x86.c
@@ -64,100 +64,100 @@ rdrand (void)
     return;
 
   /* 16-bit random numbers.  */
-  __asm__ volatile ("rdrand %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%bx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%cx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%dx;" : "=r" (number));
+  __asm__ volatile ("rdrand %%ax;": : : "%ax");
+  __asm__ volatile ("rdrand %%bx;": : : "%bx");
+  __asm__ volatile ("rdrand %%cx;": : : "%cx");
+  __asm__ volatile ("rdrand %%dx;": : : "%dx");
 
-  __asm__ volatile ("mov %%di, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%di;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%di;" : "=r" (number));
+  __asm__ volatile ("mov %%di, %%ax;\n\
+		     rdrand %%di;\n\
+		     mov %%ax, %%di;" : : : "%ax");
 
-  __asm__ volatile ("mov %%si, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%si;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%si;" : "=r" (number));
+  __asm__ volatile ("mov %%si, %%ax;\n\
+		     rdrand %%si;\n\
+		     mov %%ax, %%si;" : : : "%ax");
 
-  __asm__ volatile ("mov %%bp, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%bp;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%bp;" : "=r" (number));
+  __asm__ volatile ("mov %%bp, %%ax;\n\
+		     rdrand %%bp;\n\
+		     mov %%ax, %%bp;" : : : "%ax");
 
-  __asm__ volatile ("mov %%sp, %%ax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%sp;" : "=r" (number));
-  __asm__ volatile ("mov %%ax, %%sp;" : "=r" (number));
+  __asm__ volatile ("mov %%sp, %%ax;\n\
+		     rdrand %%sp;\n\
+		     mov %%ax, %%sp;" : : : "%ax");
 
 #ifdef __x86_64__
-  __asm__ volatile ("rdrand %%r8w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r9w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r10w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r11w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r12w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r13w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r14w;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r15w;" : "=r" (number));
+  __asm__ volatile ("rdrand %%r8w;": : : "%r8");
+  __asm__ volatile ("rdrand %%r9w;": : : "%r9");
+  __asm__ volatile ("rdrand %%r10w;": : : "%r10");
+  __asm__ volatile ("rdrand %%r11w;": : : "%r11");
+  __asm__ volatile ("rdrand %%r12w;": : : "%r12");
+  __asm__ volatile ("rdrand %%r13w;": : : "%r13");
+  __asm__ volatile ("rdrand %%r14w;": : : "%r14");
+  __asm__ volatile ("rdrand %%r15w;": : : "%r15");
 #endif
 
   /* 32-bit random numbers.  */
-  __asm__ volatile ("rdrand %%eax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%ebx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%ecx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%edx;" : "=r" (number));
+  __asm__ volatile ("rdrand %%eax;": : : "%eax");
+  __asm__ volatile ("rdrand %%ebx;": : : "%ebx");
+  __asm__ volatile ("rdrand %%ecx;": : : "%ecx");
+  __asm__ volatile ("rdrand %%edx;": : : "%edx");
 
 #ifdef __x86_64__
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%edi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%esi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%ebp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%esp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (number));
-
-  __asm__ volatile ("rdrand %%r8d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r9d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r10d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r11d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r12d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r13d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r14d;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r15d;" : "=r" (number));
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdrand %%edi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdrand %%esi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdrand %%ebp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdrand %%esp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdrand %%r8d;": : : "%r8");
+  __asm__ volatile ("rdrand %%r9d;": : : "%r9");
+  __asm__ volatile ("rdrand %%r10d;": : : "%r10");
+  __asm__ volatile ("rdrand %%r11d;": : : "%r11");
+  __asm__ volatile ("rdrand %%r12d;": : : "%r12");
+  __asm__ volatile ("rdrand %%r13d;": : : "%r13");
+  __asm__ volatile ("rdrand %%r14d;": : : "%r14");
+  __asm__ volatile ("rdrand %%r15d;": : : "%r15");
 
   /* 64-bit random numbers.  */
-  __asm__ volatile ("rdrand %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rbx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rcx;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rdx;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rdi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rsi;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rbp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (number));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (number));
-  __asm__ volatile ("rdrand %%rsp;" : "=r" (number));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (number));
-
-  __asm__ volatile ("rdrand %%r8;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r9;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r10;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r11;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r12;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r13;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r14;" : "=r" (number));
-  __asm__ volatile ("rdrand %%r15;" : "=r" (number));
+  __asm__ volatile ("rdrand %%rax;": : : "%rax");
+  __asm__ volatile ("rdrand %%rbx;": : : "%rbx");
+  __asm__ volatile ("rdrand %%rcx;": : : "%rcx");
+  __asm__ volatile ("rdrand %%rdx;": : : "%rdx");
+
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdrand %%rdi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdrand %%rsi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdrand %%rbp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdrand %%rsp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdrand %%r8;": : : "%r8");
+  __asm__ volatile ("rdrand %%r9;": : : "%r9");
+  __asm__ volatile ("rdrand %%r10;": : : "%r10");
+  __asm__ volatile ("rdrand %%r11;": : : "%r11");
+  __asm__ volatile ("rdrand %%r12;": : : "%r12");
+  __asm__ volatile ("rdrand %%r13;": : : "%r13");
+  __asm__ volatile ("rdrand %%r14;": : : "%r14");
+  __asm__ volatile ("rdrand %%r15;": : : "%r15");
 #endif
 }
 
@@ -173,100 +173,100 @@ rdseed (void)
     return;
 
   /* 16-bit random seeds.  */
-  __asm__ volatile ("rdseed %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%bx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%cx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%dx;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%ax;": : : "%ax");
+  __asm__ volatile ("rdseed %%bx;": : : "%bx");
+  __asm__ volatile ("rdseed %%cx;": : : "%cx");
+  __asm__ volatile ("rdseed %%dx;": : : "%dx");
 
-  __asm__ volatile ("mov %%di, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%di;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%di;" : "=r" (seed));
+  __asm__ volatile ("mov %%di, %%ax;\n\
+		     rdseed %%di;\n\
+		     mov %%ax, %%di;" : : : "%ax");
 
-  __asm__ volatile ("mov %%si, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%si;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%si;" : "=r" (seed));
+  __asm__ volatile ("mov %%si, %%ax;\n\
+		     rdseed %%si;\n\
+		     mov %%ax, %%si;" : : : "%ax");
 
-  __asm__ volatile ("mov %%bp, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%bp;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%bp;" : "=r" (seed));
+  __asm__ volatile ("mov %%bp, %%ax;\n\
+		     rdseed %%bp;\n\
+		     mov %%ax, %%bp;" : : : "%ax");
 
-  __asm__ volatile ("mov %%sp, %%ax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%sp;" : "=r" (seed));
-  __asm__ volatile ("mov %%ax, %%sp;" : "=r" (seed));
+  __asm__ volatile ("mov %%sp, %%ax;\n\
+		     rdseed %%sp;\n\
+		     mov %%ax, %%sp;" : : : "%ax");
 
 #ifdef __x86_64__
-  __asm__ volatile ("rdseed %%r8w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r9w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r10w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r11w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r12w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r13w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r14w;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r15w;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%r8w;": : : "%r8");
+  __asm__ volatile ("rdseed %%r9w;": : : "%r9");
+  __asm__ volatile ("rdseed %%r10w;": : : "%r10");
+  __asm__ volatile ("rdseed %%r11w;": : : "%r11");
+  __asm__ volatile ("rdseed %%r12w;": : : "%r12");
+  __asm__ volatile ("rdseed %%r13w;": : : "%r13");
+  __asm__ volatile ("rdseed %%r14w;": : : "%r14");
+  __asm__ volatile ("rdseed %%r15w;": : : "%r15");
 #endif
 
   /* 32-bit random seeds.  */
-  __asm__ volatile ("rdseed %%eax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%ebx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%ecx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%edx;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%eax;": : : "%eax");
+  __asm__ volatile ("rdseed %%ebx;": : : "%ebx");
+  __asm__ volatile ("rdseed %%ecx;": : : "%ecx");
+  __asm__ volatile ("rdseed %%edx;": : : "%edx");
 
 #ifdef __x86_64__
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%edi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%esi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%ebp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%esp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (seed));
-
-  __asm__ volatile ("rdseed %%r8d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r9d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r10d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r11d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r12d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r13d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r14d;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r15d;" : "=r" (seed));
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdseed %%edi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdseed %%esi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdseed %%ebp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdseed %%esp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdseed %%r8d;": : : "%r8");
+  __asm__ volatile ("rdseed %%r9d;": : : "%r9");
+  __asm__ volatile ("rdseed %%r10d;": : : "%r10");
+  __asm__ volatile ("rdseed %%r11d;": : : "%r11");
+  __asm__ volatile ("rdseed %%r12d;": : : "%r12");
+  __asm__ volatile ("rdseed %%r13d;": : : "%r13");
+  __asm__ volatile ("rdseed %%r14d;": : : "%r14");
+  __asm__ volatile ("rdseed %%r15d;": : : "%r15");
 
   /* 64-bit random seeds.  */
-  __asm__ volatile ("rdseed %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rbx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rcx;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rdx;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rdi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rdi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rdi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsi, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rsi;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsi;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rbp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rbp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rbp;" : "=r" (seed));
-
-  __asm__ volatile ("mov %%rsp, %%rax;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%rsp;" : "=r" (seed));
-  __asm__ volatile ("mov %%rax, %%rsp;" : "=r" (seed));
-
-  __asm__ volatile ("rdseed %%r8;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r9;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r10;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r11;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r12;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r13;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r14;" : "=r" (seed));
-  __asm__ volatile ("rdseed %%r15;" : "=r" (seed));
+  __asm__ volatile ("rdseed %%rax;": : : "%rax");
+  __asm__ volatile ("rdseed %%rbx;": : : "%rbx");
+  __asm__ volatile ("rdseed %%rcx;": : : "%rcx");
+  __asm__ volatile ("rdseed %%rdx;": : : "%rdx");
+
+  __asm__ volatile ("mov %%rdi, %%rax;\n\
+		     rdseed %%rdi;\n\
+		     mov %%rax, %%rdi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsi, %%rax;\n\
+		     rdseed %%rsi;\n\
+		     mov %%rax, %%rsi;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rbp, %%rax;\n\
+		     rdseed %%rbp;\n\
+		     mov %%rax, %%rbp;" : : : "%rax");
+
+  __asm__ volatile ("mov %%rsp, %%rax;\n\
+		     rdseed %%rsp;\n\
+		     mov %%rax, %%rsp;" : : : "%rax");
+
+  __asm__ volatile ("rdseed %%r8;": : : "%r8");
+  __asm__ volatile ("rdseed %%r9;": : : "%r9");
+  __asm__ volatile ("rdseed %%r10;": : : "%r10");
+  __asm__ volatile ("rdseed %%r11;": : : "%r11");
+  __asm__ volatile ("rdseed %%r12;": : : "%r12");
+  __asm__ volatile ("rdseed %%r13;": : : "%r13");
+  __asm__ volatile ("rdseed %%r14;": : : "%r14");
+  __asm__ volatile ("rdseed %%r15;": : : "%r15");
 #endif
 }
 
diff --git a/gdb/testsuite/gdb.reverse/insn-reverse.exp b/gdb/testsuite/gdb.reverse/insn-reverse.exp
index 1a575b2d43e..206e765dd50 100644
--- a/gdb/testsuite/gdb.reverse/insn-reverse.exp
+++ b/gdb/testsuite/gdb.reverse/insn-reverse.exp
@@ -95,7 +95,9 @@ proc test { func testcase_nr } {
 
 	    if { $prev_insn_addr == $insn_addr } {
 		# Failed to make progress, might have run into SIGILL.
-		unsupported "no progress at: $expect_out(2,string)"
+		fail "no progress at: $expect_out(2,string)"
+		# Ignore the last instruction recorded
+		incr count -1
 		break
 	    }
 
-- 
2.41.0

Re: [PATCH v2 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Tom Tromey]

>>>>> "Bruno" == Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:

Bruno> It also adds a new parameter to get_hexadecimal_valueof, so that we can
Bruno> use it without generating new passes, otherwise we'd get multiple
Bruno> duplicate test names. This change shouldn't affect any other test using
Bruno> this proc.

You can just pass different test names instead.

Bruno> +proc get_current_pc {} {
Bruno> +    set pc 0
Bruno> +    gdb_test_multiple "print \$pc" "" {
Bruno> +	-re -wrap ".*0x(\[0-9a-f\]+).*" {
Bruno> +	    set pc $expect_out(1,string)
Bruno> +	}
Bruno> +    }
Bruno> +    return $pc

It seems to me that this shouldn't be needed.

Bruno> diff --git a/gdb/testsuite/gdb.reverse/step-reverse.exp b/gdb/testsuite/gdb.reverse/step-reverse.exp
Bruno> index 4b78a8f8fb7..9ff97bfde42 100644
Bruno> --- a/gdb/testsuite/gdb.reverse/step-reverse.exp
Bruno> +++ b/gdb/testsuite/gdb.reverse/step-reverse.exp
Bruno> @@ -28,6 +28,16 @@ if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
Bruno>      return -1
Bruno>  }
 
Bruno> +proc get_current_pc {} {
Bruno> +    set pc 0
Bruno> +    gdb_test_multiple "print \$pc" "" {
Bruno> +	-re -wrap ".*0x(\[0-9a-f\]+).*" {
Bruno> +	    set pc $expect_out(1,string)
Bruno> +	}
Bruno> +    }
Bruno> +    return $pc

Same with this one.

Tom

Re: [PATCH v2 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Bruno Larsen]

On 28/07/2023 15:14, Tom Tromey wrote:
>>>>>> "Bruno" == Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:
> Bruno> It also adds a new parameter to get_hexadecimal_valueof, so that we can
> Bruno> use it without generating new passes, otherwise we'd get multiple
> Bruno> duplicate test names. This change shouldn't affect any other test using
> Bruno> this proc.
>
> You can just pass different test names instead.

I mean, yeah I can, but since it is in a loop, the differences would 
only be a counter at the end of the test case. Are we really getting any 
value from that? To me it seems like it would just boggle down the sum 
file with meaningless "tests" that aren't exercising any relevant code 
paths. I can do it if you disagree, though, it isn't a big deal.

One thing I did think of was that I should test for default value when 
getting the PC, to make sure that we didnt get a valid PC at first, then 
started getting "no registers" or similar. I'll have that on v3

-- 
Cheers,
Bruno

>
> Bruno> +proc get_current_pc {} {
> Bruno> +    set pc 0
> Bruno> +    gdb_test_multiple "print \$pc" "" {
> Bruno> +	-re -wrap ".*0x(\[0-9a-f\]+).*" {
> Bruno> +	    set pc $expect_out(1,string)
> Bruno> +	}
> Bruno> +    }
> Bruno> +    return $pc
>
> It seems to me that this shouldn't be needed.
>
> Bruno> diff --git a/gdb/testsuite/gdb.reverse/step-reverse.exp b/gdb/testsuite/gdb.reverse/step-reverse.exp
> Bruno> index 4b78a8f8fb7..9ff97bfde42 100644
> Bruno> --- a/gdb/testsuite/gdb.reverse/step-reverse.exp
> Bruno> +++ b/gdb/testsuite/gdb.reverse/step-reverse.exp
> Bruno> @@ -28,6 +28,16 @@ if { [prepare_for_testing "failed to prepare" $testfile $srcfile] } {
> Bruno>      return -1
> Bruno>  }
>   
> Bruno> +proc get_current_pc {} {
> Bruno> +    set pc 0
> Bruno> +    gdb_test_multiple "print \$pc" "" {
> Bruno> +	-re -wrap ".*0x(\[0-9a-f\]+).*" {
> Bruno> +	    set pc $expect_out(1,string)
> Bruno> +	}
> Bruno> +    }
> Bruno> +    return $pc
>
> Same with this one.
>
> Tom
>

Re: [PATCH v2 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Tom Tromey]

>>>>> "Bruno" == Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:

Bruno> I mean, yeah I can, but since it is in a loop, the differences would
Bruno> only be a counter at the end of the test case. Are we really getting
Bruno> any value from that? To me it seems like it would just boggle down the
Bruno> sum file with meaningless "tests" that aren't exercising any relevant
Bruno> code paths. I can do it if you disagree, though, it isn't a big deal.

Yeah, I don't know if it's useful -- but are any of these passes useful?
The problem I think is that the proc can unconditionally emit a fail, so
having a corresponding pass seems to make sense.

Tom

Re: [PATCH v2 3/4] gdb/testsuite: fix testing gdb.reverse/step-reverse.exp with clang

[Bruno Larsen]

On 28/07/2023 16:18, Tom Tromey wrote:
>>>>>> "Bruno" == Bruno Larsen via Gdb-patches <gdb-patches@sourceware.org> writes:
> Bruno> I mean, yeah I can, but since it is in a loop, the differences would
> Bruno> only be a counter at the end of the test case. Are we really getting
> Bruno> any value from that? To me it seems like it would just boggle down the
> Bruno> sum file with meaningless "tests" that aren't exercising any relevant
> Bruno> code paths. I can do it if you disagree, though, it isn't a big deal.
>
> Yeah, I don't know if it's useful -- but are any of these passes useful?
> The problem I think is that the proc can unconditionally emit a fail, so
> having a corresponding pass seems to make sense.
>
Ah, yeah, in that angle it does make sense. I'll do it for v3!

-- 
Cheers,
Bruno