Re: [committed][gdb/testsuite] Add gdb.base/eh_return.exp

[Tom de Vries <tdevries@suse.de>]

[-- Attachment #1: Type: text/plain, Size: 4048 bytes --]

On 8/27/20 1:14 AM, Tom de Vries wrote:
> On 8/26/20 8:02 PM, Tom de Vries wrote:
>> On 8/26/20 7:20 PM, Luis Machado wrote:
>>> On 8/16/20 8:31 AM, Tom de Vries wrote:
>>>> Hi,
>>>>
>>>> In PR25350, an internal error was reported:
>>>> ...
>>>> (gdb) break *eh2+0x7e
>>>> Breakpoint 1 at 0x13e2: file small.c, line 38.
>>>> (gdb) run
>>>> Starting program: a.out
>>>> [Thread debugging using libthread_db enabled]
>>>> Using host libthread_db library "/usr/lib/libthread_db.so.1".
>>>>
>>>> Breakpoint 1, 0x00005555555553e2 in eh2 (
>>>> frame.c:558: internal-error: frame_id get_frame_id(frame_info*): \
>>>>    Assertion `stashed' failed.
>>>> A problem internal to GDB has been detected,
>>>> further debugging may prove unreliable.
>>>> ...
>>>>
>>>> The internal error does not reproduce after recent commit 547ce8f00b
>>>> "[gdb/backtrace] Fix printing of fortran string args".
>>>>
>>>> Add the corresponding test-case as regression test, given that the
>>>> code is
>>>> rather atypical.
>>>>
>>>> Tested on x86_64-linux.
>>>
>>> This doesn't work so well on aarch64-linux because the breakpoint
>>> location from before running the program doesn't exists after running
>>> it. Things get relocated.
>>>
>>
>> Right, I don't think that's architecture-related.
>>
>> I'm running into the same issue on x86_64-linux with target board
>> unix/-fPIE/-pie.
>>
>>> -- 
>>>
>>> (gdb) break *0x000000000000099c^M
>>> Breakpoint 1 at 0x99c: file
>>> /home/luis.machado/work/tcwg/build/binutils-gdb-master/gdb/testsuite/../../../../repos/binutils-gdb/gdb/testsuite/gdb.base/eh_return.c,
>>> line 54.^M
>>> (gdb) PASS: gdb.base/eh_return.exp: setting breakpoint at
>>> *0x000000000000099c
>>> run ^M
>>> Starting program:
>>> /home/luis.machado/work/tcwg/build/binutils-gdb-master/gdb/testsuite/outputs/gdb.base/eh_return/eh_return
>>> ^M
>>> Warning:^M
>>> Cannot insert breakpoint 1.^M
>>> Cannot access memory at address 0x99c^M
>>> ^M
>>> (gdb) FAIL: gdb.base/eh_return.exp: hit breakpoint
>>>
>>> -- 
>>>
>>> Shouldn't we insert a breakpoint after the relocation has happened?
>>
>> As I mentioned in the test-case:
>> ...
>> # The assert did not reproduce when running to main, and continuing to
>> # the breakpoint, so instead, run to the breakpoint.
>> ...
>> So I'm afraid that's not going to work.
>>
>>> Or
>>> add the breakpoint to a symbolic reference?
>>
>> Yeah, that might work.  If not, we'll have to restrict this to no-pie.
> 
> So, I tried this using:
> ...
> $ git diff
> diff --git a/gdb/testsuite/gdb.base/eh_return.c
> b/gdb/testsuite/gdb.base/eh_ret
> urn.c
> index 80eea9c783..ba8c1c51aa 100644
> --- a/gdb/testsuite/gdb.base/eh_return.c
> +++ b/gdb/testsuite/gdb.base/eh_return.c
> @@ -51,6 +51,8 @@ eh2 (void *p)
>  {
>    eh2a (val, val, val, val, val, val, val, val, p);
>    __builtin_eh_return (0, p);
> + here:
> +  (void)0;
>  }
> 
> 
> diff --git a/gdb/testsuite/gdb.base/eh_return.exp
> b/gdb/testsuite/gdb.base/eh_r
> eturn.exp
> index e38398ed37..182b2ed86d 100644
> --- a/gdb/testsuite/gdb.base/eh_return.exp
> +++ b/gdb/testsuite/gdb.base/eh_return.exp
> @@ -22,23 +22,9 @@ if {[prepare_for_testing "failed to prepare"
>  clean_restart ${binfile}
> 
> -gdb_breakpoint "*$address" message
> +gdb_breakpoint "eh2:here" message
> 
>  # The assert did not reproduce when running to main, and continuing to the
>  # breakpoint, so instead, run to the breakpoint.
> ...
> 
> But then we run into another internal-error while setting the
> breakpoint, see PR25067:
> ...
> src/gdb/linespec.c:3233: internal-error: void decode_line_full(const
> event_location*, int, program_space*, symtab*, int, linespec_result*,
> const char*, const char*): Assertion `result.size () == 1 ||
> canonical->pre_expanded' failed.^M
> A problem internal to GDB has been detected,^M
> further debugging may prove unreliable.^M
> ...
> 

Something I didn't realize when trying this out: __builtin_eh_return is
a noreturn builtin, so adding a label after it doesn't work.

So I've gone with nopie.

Committed as attached.

Thanks,
- Tom

[-- Attachment #2: 0001-gdb-testsuite-Add-nopie-to-gdb.base-eh_return.exp.patch --]
[-- Type: text/x-patch, Size: 1269 bytes --]

[gdb/testsuite] Add nopie to gdb.base/eh_return.exp

When running test-case gdb.base/eh_return.exp with target board
unix/-fPIE/-pie, we run into:
...
(gdb) break *0x88e^M
Breakpoint 1 at 0x88e: file eh_return.c, line 54.^M
(gdb) PASS: gdb.base/eh_return.exp: setting breakpoint at *0x88e
run ^M
Starting program: eh_return ^M
Warning:^M
Cannot insert breakpoint 1.^M
Cannot access memory at address 0x88e^M
^M
(gdb) FAIL: gdb.base/eh_return.exp: hit breakpoint
...

The problem is that gdb does not support setting breakpoints on unrelocated
addresses.

Fix this by using nopie for the test-case.

Tested on x86_64-linux.

gdb/testsuite/ChangeLog:

2020-08-31  Tom de Vries  <tdevries@suse.de>

	* gdb.base/eh_return.exp: Use nopie.

---
 gdb/testsuite/gdb.base/eh_return.exp | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/gdb/testsuite/gdb.base/eh_return.exp b/gdb/testsuite/gdb.base/eh_return.exp
index e38398ed37..78a4af4b0c 100644
--- a/gdb/testsuite/gdb.base/eh_return.exp
+++ b/gdb/testsuite/gdb.base/eh_return.exp
@@ -18,7 +18,8 @@
 
 standard_testfile
 
-if {[prepare_for_testing "failed to prepare" $testfile $srcfile debug]} {
+if {[prepare_for_testing "failed to prepare" $testfile $srcfile \
+	 {debug nopie}]} {
     return -1
 }