Date: Wed, 23 Jan 2019 10:10:43 -0800 From: Mark Millard <marklmi@yahoo.com> To: John Baldwin <jhb@FreeBSD.org> Cc: "freebsd-ppc@freebsd.org" <freebsd-ppc@FreeBSD.org> Subject: Re: GDB TLS testing [actually running some tests finally: a success with -pthread used] Message-ID: <20B92DA0-33B7-44D1-AB92-E3DD55A8B7CE@yahoo.com> In-Reply-To: <DBF4C75D-213E-42E2-8682-6B3078FCB8D5@yahoo.com> References: <b0f5b62f-54fb-309f-a578-7b4d9e340a55@FreeBSD.org> <19343397-859C-4629-A4A5-B0DCDE25957B@yahoo.com> <AB56B3D1-1762-4115-B7B4-91D4B997F1C4@yahoo.com> <eff896f3-e385-9a12-5132-3884e69689ca@FreeBSD.org> <D0B83B01-5428-42F0-B785-221826AF7EC4@yahoo.com> <a1da2471-535d-1497-ddf3-93aa0d29df59@FreeBSD.org> <5AA68ED2-2615-438B-A6AE-406CBD8E49F7@yahoo.com> <20027C29-0093-4001-A135-23783F8B87F3@yahoo.com> <4048D2A4-7E14-481C-9B5D-00567BCF4463@yahoo.com> <2AAC9738-73BD-475A-888A-252EE853A5C6@yahoo.com> <493AC0BE-3EC6-42B7-B027-FFB6454761B5@yahoo.com> <B07EC3F9-6C08-4B5D-89D0-4048AF305A2D@yahoo.com> <52E66D9B-C332-4565-B8E7-F54F6454B062@yahoo.com> <a0e02258-3e43-fb91-566d-29b48fe81e77@FreeBSD.org> <DBF4C75D-213E-42E2-8682-6B3078FCB8D5@yahoo.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 2019-Jan-23, at 09:59, Mark Millard <marklmi at yahoo.com> wrote:
> On 2019-Jan-23, at 09:02, John Baldwin <jhb at FreeBSD.org> wrote:
>=20
>> On 1/23/19 12:19 AM, Mark Millard wrote:
>>>=20
>>>=20
>>> On 2019-Jan-22, at 22:53, Mark Millard <marklmi@yahoo.com> wrote:
>>>=20
>>>=20
>>>=20
>>>> On 2019-Jan-22, at 19:19, Mark Millard <marklmi at yahoo.com> =
wrote:
>>>>=20
>>>>=20
>>>>=20
>>>>> On 2019-Jan-22, at 18:32, Mark Millard <marklm at yahoo.com> =
wrote:
>>>>>=20
>>>>>=20
>>>>>=20
>>>>>> On 2019-Jan-22, at 17:06, Mark Millard <marklmi at yahoo.com> =
wrote:
>>>>>>=20
>>>>>>=20
>>>>>>=20
>>>>>> . . .
>>>>>> So I'm trying:
>>>>>>=20
>>>>>> # git clean -f
>>>>>> # rm */config.cache */*/config.cache
>>>>>> # env CPATH=3D/usr/local/include ./configure
>>>>>> . . .
>>>>>> # env CPATH=3D/usr/local/include gmake
>>>>>> . . .
>>>>>>=20
>>>>>> in order to try to add paths after the command line -I paths.
>>>>>>=20
>>>>>> . . . This looks like it built. I've not used the build yet.
>>>>>>=20
>>>>>=20
>>>>> Looking at a *.core did not go well for my context:
>>>>>=20
>>>>> # cc -g -O2 tls_gdb_test.c=20
>>>>> tls_gdb_test.c:16:2: warning: indirection of non-volatile null =
pointer will be deleted, not trap [-Wnull-dereference]
>>>>> *(char *)NULL =3D 1;
>>>>> ^~~~~~~~~~~~~
>>>>> tls_gdb_test.c:16:2: note: consider using __builtin_trap() or =
qualifying pointer with 'volatile'
>>>>> 1 warning generated.
>>=20
>> For me, builtin_trap didn't DTRT on ppc before, but that was probably =
using
>> gcc rather than clang (and possibly using gcc4.2).
>>=20
>>>>> . . .
>>>>=20
>>>>=20
>>>> So far all tried-combinations of using gcc versions for build
>>>> the test program and/or building the gdb used do not work for
>>>> "p id" and "p &id" in doing the test. clang is not essential
>>>> to the behavior observed.
>>>=20
>>> Using "-g -O2 -pthread" to build the test program via system
>>> clang or gcc8 (for example) did lead to the likes of:
>>>=20
>>> (gdb) run=20
>>> Starting program: /root/c_tests/a.out=20
>>> main: PID 15350
>>> id =3D 15350 (0x810055020)
>>>=20
>>> ^C
>>> Program received signal SIGINT, Interrupt.
>>> main (ac=3D<optimized out>, av=3D<optimized out>) at =
tls_gdb_test.c:16
>>> 16 *(char *)NULL =3D 1;
>>> (gdb) bt
>>> #0 main (ac=3D<optimized out>, av=3D<optimized out>) at =
tls_gdb_test.c:16
>>> (gdb) info threads
>>> Id Target Id Frame=20
>>> * 1 LWP 100324 of process 15350 main (ac=3D<optimized out>, =
av=3D<optimized out>) at tls_gdb_test.c:16
>>> (gdb) p id
>>> $1 =3D 15350
>>> (gdb) p &id
>>> $2 =3D (int *) 0x810055020
>>>=20
>>> So it appears one branch of:
>>>=20
>>> static void
>>> fbsd_fetch_rtld_offsets (struct gdbarch *gdbarch, struct =
fbsd_pspace_data *data)
>>> {
>>> TRY
>>> {
>>> /* Fetch offsets from debug symbols in rtld. */
>>> data->off_linkmap =3D parse_and_eval_long ("&((Obj_Entry =
*)0)->linkmap");
>>> data->off_tlsindex =3D parse_and_eval_long ("&((Obj_Entry =
*)0)->tlsindex");
>>> data->rtld_offsets_valid =3D true;
>>> return;
>>> }
>>> CATCH (e, RETURN_MASK_ERROR)
>>> {
>>> data->off_linkmap =3D -1;
>>> }
>>> END_CATCH
>>>=20
>>> TRY
>>> {
>>> /* Fetch offsets from global variables in libthr. Note that
>>> this does not work for single-threaded processes that are not
>>> linked against libthr. */
>>> data->off_linkmap =3D fbsd_read_integer_by_name(gdbarch,
>>> =
"_thread_off_linkmap");
>>> data->off_tlsindex =3D fbsd_read_integer_by_name(gdbarch,
>>> =
"_thread_off_tlsindex");
>>> data->rtld_offsets_valid =3D true;
>>> return;
>>> }
>>> CATCH (e, RETURN_MASK_ERROR)
>>> {
>>> data->off_linkmap =3D -1;
>>> }
>>> END_CATCH
>>> }
>>>=20
>>> is working when -pthread is used.
>>=20
>> Yes. The second one only works for programs linked against -lthr. =
Otherwise
>> you need to have built your system with debug symbols (which is the =
default),
>> and gdb needs to be able to access =
/usr/lib/debug/libexec/ld-elf.so.1.debug
>> to determine the offsets of the two fields in Obj_Entry (this is what =
the
>> first TRY clause does).
>=20
> I buildworld buildkernel with debug symbols for both and install them:
>=20
> # ls -lT /usr/lib/debug/libexec/ld-elf.so.1.debug
> -r--r--r-- 1 root wheel 576344 Dec 11 22:58:11 2018 =
/usr/lib/debug/libexec/ld-elf.so.1.debug
>=20
> # ls -lT /usr/libexec/ld-elf.so.1 /libexec/ld-elf.so.1
> -r-xr-xr-x 1 root wheel 184400 Dec 9 02:35:05 2018 =
/libexec/ld-elf.so.1
> lrwxr-xr-x 1 root wheel 25 Dec 11 22:58:12 2018 =
/usr/libexec/ld-elf.so.1 -> ../../libexec/ld-elf.so.1
>=20
> So I think the first TRY clause does not work.
>=20
> In the tested gdb used on the a.out I'm testing I get:
>=20
> (gdb) p &((Obj_Entry *)0)->linkmap
> No symbol "Obj_Entry" in current context.
>=20
> gdb does not report reading symbols from or for:
>=20
> /usr/lib/debug/libexec/ld-elf.so.1.debug
> or:
> /usr/libexec/ld-elf.so.1
> or:
> /libexec/ld-elf.so.1
>=20
> It only reports reading them from/for the a.out .
It neded up that I had a littel time so . . .
I tried /usr/local/bin/gdb and for it:
(gdb) p &((Obj_Entry *)0)->linkmap
$1 =3D (struct link_map *) 0x238
So the lack of finding Obj_Entry via the test gdb seems to be
specific to the test gdb, not a problem for devel/gdb .
May be the test gdb has some sort of build problem in my context,
given that I used CPATH to get things to build?
=3D=3D=3D
Mark Millard
marklmi at yahoo.com
( dsl-only.net went
away in early 2018-Mar)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20B92DA0-33B7-44D1-AB92-E3DD55A8B7CE>