Re: [PATCH v3 2/2] Make gdbserver work with filename-only binaries

[Simon Marchi <simon.marchi@ericsson.com>]

On 2018-02-27 10:58 PM, Sergio Durigan Junior wrote:
> On Tuesday, February 27 2018, I wrote:
> 
>> Simon mentioned on IRC that, after the startup-with-shell feature has
>> been implemented on gdbserver, it is not possible to specify a
>> filename-only binary, like:
>>
>>   $ gdbserver :1234 a.out
>>   /bin/bash: line 0: exec: a.out: not found
>>   During startup program exited with code 127.
>>   Exiting
>>
>> This happens on systems where the current directory "." is not listed
>> in the PATH environment variable.  Although including "." in the PATH
>> variable is a possible workaround, this can be considered a regression
>> because before startup-with-shell it was possible to use only the
>> filename (due to reason that gdbserver used "exec*" directly).
>>
>> The idea of the patch is to verify if the program path provided by the
>> user (or by the remote protocol) contains a directory separator
>> character.  If it doesn't, it means we're dealing with a filename-only
>> binary, so we call "gdb_abspath" to properly expand it and transform
>> it into a full path.  Otherwise, we leave the program path untouched.
>> This mimicks the behaviour seen on GDB (look at "openp" and
>> "attach_inferior", for example).
>>
>> I am also submitting a testcase which exercises the scenario described
>> above.  This test requires gdbserver to be executed in a different CWD
>> than the original, so I also created a helper function, "with_cwd" (on
>> testsuite/lib/gdb.exp), which takes care of cd'ing into and out of the
>> specified dir.
> 
> This part is still giving me a few headaches.  I've just noticed that
> two builders reported the new test as FAIL.  When I run it here, I can't
> reproduce it, which makes me wonder that it's racy.  I don't know if the
> culprit is the new "with_cwd" logic or not.  Curiously, both builders
> reporting the failure are running on s390x
> (Debian-s390x-native-extended-gdbserver-m64 and RHEL-s390x-m64).
> 
> Here's an excerpt of report from one of them,
> Debian-s390x-native-extended-gdbserver-m64:
> 
> ----------------
> (gdb) file /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/outputs/gdb.server/abspath/abspath
> Reading symbols from /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/outputs/gdb.server/abspath/abspath...done.
> (gdb) set remote exec-file /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/outputs/gdb.server/abspath/abspath
> (gdb) set remote exec-file /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/outputs/gdb.server/abspath/abspath
> (gdb) disconnect
> Ending remote debugging.
> (gdb) PASS: gdb.server/abspath.exp: disconnect
> Switching to directory /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/outputs/gdb.server/abspath (saved CWD: /home/dje/debian-jessie-s390x-1/debian
> -s390x-native-extended-gdbserver/build/gdb/testsuite).
> spawn /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/../gdbserver/gdbserver --once :2347 abspath
> Can't bind address: Address already in use.
> Exiting
> Port 2347 is already in use.
> spawn /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/../gdbserver/gdbserver --once :2348 abspath
> Process /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/outputs/gdb.server/abspath/abspath created; pid = 21406
> Listening on port 2348
> target extended-remote localhost:2348
> Remote debugging using localhost:2348
> Reading /lib/ld64.so.1 from remote target...
> warning: File transfers from remote targets can be slow. Use "set sysroot" to access files locally instead.
> Reading /lib/ld64.so.1 from remote target...
> Reading symbols from target:/lib/ld64.so.1...Reading /lib/ld-2.19.so from remote target...
> Reading /lib/.debug/ld-2.19.so from remote target...
> (no debugging symbols found)...done.
> 0x000003fffdfd9280 in ?? () from target:/lib/ld64.so.1
> Protocol error: qXfer:btrace-conf (read-btrace-conf) conflicting enabled responses.
> (gdb) break main
> Breakpoint 1 at 0x800005bc: file /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite/../../../binutils-gdb/gdb/testsuite/gdb.server/normal.c, line 23.
> (gdb) continue
> The program is not being run.
> (gdb) FAIL: gdb.server/abspath.exp: continue to main (the program is no longer running)
> Switching back to /home/dje/debian-jessie-s390x-1/debian-s390x-native-extended-gdbserver/build/gdb/testsuite.
> ----------------
> 
> They're both running the testsuite in parallel mode (-j8), but when I
> run it locally in parallel I can't reproduce (even when I run in a
> loop).
> 
> Anyway, I'll try to find out why this happens.  I don't see how changing
> the CWD in a test could impact its results, so I'm guessing there's
> something else at play here.
> 
> Thanks,
> 

Maybe it's just because of this issue?  Note the btrace-related
error message.

https://sourceware.org/ml/gdb-patches/2018-02/msg00352.html