Date: Fri, 16 Jul 2010 12:51:38 +0300 From: "Reko Turja" <reko.turja@liukuma.net> To: "Jeremy Chadwick" <freebsd@jdc.parodius.com> Cc: "Mikhail T." <mi+thun@aldan.algebra.com>, freebsd-stable@freebsd.org, Henrik /KaarPoSoft <henrik@kaarposoft.dk> Subject: Re: openldap client GSSAPI authentication segfaults in fbsd8stablei386 Message-ID: <94382A43AE5842FD9571AE05688F09D8@rivendell> In-Reply-To: <20100716092209.GA99001@icarus.home.lan> References: <4C3CC831.7040005@kaarposoft.dk> <20100713210729.GA11943@icarus.home.lan> <0228E401B70A4023A6F86A2ADAE59EF9@rivendell> <008D0251AE4F4A2DBAA1369410565B61@rivendell> <20100715162251.GA73929@icarus.home.lan> <20100716083617.GA97981@icarus.home.lan> <3FE6787E5CAC4C108C031CA6C8044FE4@rivendell> <20100716092209.GA99001@icarus.home.lan>
next in thread | previous in thread | raw e-mail | index | archive | help
>> void
>> 232 _gss_mg_error(struct _gss_mech_switch *m, OM_uint32 maj,
>> OM_uint32 min)
>> 233 {
>> 234 OM_uint32 major_status, minor_status;
>> 235 OM_uint32 message_content;
>> 236 struct mg_thread_ctx *mg;
>> 237
>> 238 mg =3D &last_error_context;
>> 239
>> 240 gss_release_buffer(&minor_status, &mg->maj_error);
>> 241 gss_release_buffer(&minor_status, &mg->min_error);
>> 242
>> 243 mg->mech =3D &m->gm_mech_oid;
>> 244 mg->maj_stat =3D maj;
>>
>> when I give following comands, gdb tells me:
>>
>> (gdb) p last_error_context
>> Cannot find thread-local variables on this target
>> (gdb) p &last_error_context
>> Cannot find thread-local variables on this target
>> (gdb) p mg
>> No symbol "mg" in current context.
>> (gdb)
>
> I'm not sure if you're familiar with C or not.
>
> This is because gdb's context is at the wrong frame. In the=20
> backtrace
> you provided originally, you'd need to do:
>
> (gdb) frame 2
>
> To look at the variables associated with gss_display_status.c.
>
> last_error_context could be an exported variable (you'd need to look
> through the source to find out where it's declared), so you might=20
> have
> to print it with its source filename referenced. The print command=20
> I
> gave you before (p/x filename.c::variable) didn't work, and that's a
> surprise since the gdb documentation I read says it should.
>
> Also be aware that mg is a struct, so "p mg" won't tell you much,=20
> other
> than whether or not it's null. You're probably more interested in
> members of the struct, such as mg->maj_error and mg->min_error, and
> other struct members.
I gave f 2 before entering the commands above, and unless I'm much=20
mistaken, 'p mg' should at least give the pointer to the start of the=20
struct in question, so 'No symbol "mg" in current context' is mildly=20
interesting reply from GDB :) If I understand the code in question=20
right the last_error_context's address should be copied to mg for=20
accessing the error structure defined elsewhere in the scope of the=20
while.
the definition is in same file and is as follows:
176 #if defined(__sparc64__) || defined(__arm__) ||=20
defined(__mips__)
177
178 /*
179 * These platforms don't support TLS on FreeBSD - threads will=20
just
180 * have to step on each other's error values for now.
181 */
182 #define __thread
183
184 #endif
185
186 struct mg_thread_ctx {
187 gss_OID mech;
188 OM_uint32 maj_stat;
189 OM_uint32 min_stat;
190 gss_buffer_desc maj_error;
191 gss_buffer_desc min_error;
192 };
193 static __thread struct mg_thread_ctx last_error_context
-Reko=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?94382A43AE5842FD9571AE05688F09D8>