* Re: [RFA] Fix for searching for data symbols
2002-01-19 9:33 [RFA] Fix for searching for data symbols Daniel Jacobowitz
@ 2002-01-19 9:56 ` Michael Snyder
2002-01-19 10:05 ` Daniel Jacobowitz
2002-01-29 21:56 ` Daniel Jacobowitz
2002-01-30 15:26 ` Elena Zannoni
2 siblings, 1 reply; 7+ messages in thread
From: Michael Snyder @ 2002-01-19 9:56 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: gdb-patches
Daniel Jacobowitz wrote:
>
> This patch fixes the problem I described in:
> http://sources.redhat.com/ml/gdb/2002-01/msg00223.html
>
> Basically, since we use the bounds of the text segment to search for symbols
> in find_pc_*, don't let them return a bogus answer for a symbol we know is
> off in data land.
>
> OK to commit, or does anyone prefer a different approach to this?
Is there any chance that a data symbol that we don't want
would shadow a text symbol that we do want? Eg. maybe in
one module we have
static int foo;
and in another we have
extern int foo(void);
Maybe they are even in separate objfiles (shared libs).
Maybe lookup_minimal_symbol will find the wrong one first,
and your change will terminate the search before we would
have found the right one?
Just speculating...
>
> --
> Daniel Jacobowitz Carnegie Mellon University
> MontaVista Software Debian GNU/Linux Developer
>
> 2002-01-19 Daniel Jacobowitz <drow@mvista.com>
>
> * symtab.c (find_pc_sect_psymtab): Do not search psymtabs for
> data symbols, since we search based on textlow and texthigh.
> (find_pc_sect_symtab): Likewise.
>
> Index: symtab.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/symtab.c,v
> retrieving revision 1.51
> diff -u -p -r1.51 symtab.c
> --- symtab.c 2001/12/21 22:32:37 1.51
> +++ symtab.c 2002/01/19 17:29:09
> @@ -354,12 +354,22 @@ find_pc_sect_psymtab (CORE_ADDR pc, asec
> {
> register struct partial_symtab *pst;
> register struct objfile *objfile;
> + struct minimal_symbol *msymbol;
>
> + /* If we know that this is not a text address, return failure. */
> + msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> + if (msymbol
> + && (msymbol->type == mst_data
> + || msymbol->type == mst_bss
> + || msymbol->type == mst_abs
> + || msymbol->type == mst_file_data
> + || msymbol->type == mst_file_bss))
> + return NULL;
> +
> ALL_PSYMTABS (objfile, pst)
> {
> if (pc >= pst->textlow && pc < pst->texthigh)
> {
> - struct minimal_symbol *msymbol;
> struct partial_symtab *tpst;
>
> /* An objfile that has its functions reordered might have
> @@ -370,7 +380,6 @@ find_pc_sect_psymtab (CORE_ADDR pc, asec
> section == 0) /* can't validate section this way */
> return (pst);
>
> - msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> if (msymbol == NULL)
> return (pst);
>
> @@ -1385,6 +1394,17 @@ find_pc_sect_symtab (CORE_ADDR pc, asect
> register struct partial_symtab *ps;
> register struct objfile *objfile;
> CORE_ADDR distance = 0;
> + struct minimal_symbol *msymbol;
> +
> + /* If we know that this is not a text address, return failure. */
> + msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> + if (msymbol
> + && (msymbol->type == mst_data
> + || msymbol->type == mst_bss
> + || msymbol->type == mst_abs
> + || msymbol->type == mst_file_data
> + || msymbol->type == mst_file_bss))
> + return NULL;
>
> /* Search all symtabs for the one whose file contains our address, and which
> is the smallest of all the ones containing the address. This is designed
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [RFA] Fix for searching for data symbols
2002-01-19 9:56 ` Michael Snyder
@ 2002-01-19 10:05 ` Daniel Jacobowitz
2002-01-19 11:26 ` Michael Snyder
0 siblings, 1 reply; 7+ messages in thread
From: Daniel Jacobowitz @ 2002-01-19 10:05 UTC (permalink / raw)
To: Michael Snyder; +Cc: gdb-patches
On Sat, Jan 19, 2002 at 09:52:22AM -0800, Michael Snyder wrote:
> Daniel Jacobowitz wrote:
> >
> > This patch fixes the problem I described in:
> > http://sources.redhat.com/ml/gdb/2002-01/msg00223.html
> >
> > Basically, since we use the bounds of the text segment to search for symbols
> > in find_pc_*, don't let them return a bogus answer for a symbol we know is
> > off in data land.
> >
> > OK to commit, or does anyone prefer a different approach to this?
>
> Is there any chance that a data symbol that we don't want
> would shadow a text symbol that we do want? Eg. maybe in
> one module we have
> static int foo;
> and in another we have
> extern int foo(void);
> Maybe they are even in separate objfiles (shared libs).
> Maybe lookup_minimal_symbol will find the wrong one first,
> and your change will terminate the search before we would
> have found the right one?
>
> Just speculating...
Well, we aren't using lookup_minimal_symbol; we're using
lookup_minimal_symbol_by_pc_section. I don't think it's possible for
it to get the wrong one, since it's by address instead of by name.
--
Daniel Jacobowitz Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFA] Fix for searching for data symbols
2002-01-19 10:05 ` Daniel Jacobowitz
@ 2002-01-19 11:26 ` Michael Snyder
0 siblings, 0 replies; 7+ messages in thread
From: Michael Snyder @ 2002-01-19 11:26 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: gdb-patches
Daniel Jacobowitz wrote:
>
> On Sat, Jan 19, 2002 at 09:52:22AM -0800, Michael Snyder wrote:
> > Daniel Jacobowitz wrote:
> > >
> > > This patch fixes the problem I described in:
> > > http://sources.redhat.com/ml/gdb/2002-01/msg00223.html
> > >
> > > Basically, since we use the bounds of the text segment to search for symbols
> > > in find_pc_*, don't let them return a bogus answer for a symbol we know is
> > > off in data land.
> > >
> > > OK to commit, or does anyone prefer a different approach to this?
> >
> > Is there any chance that a data symbol that we don't want
> > would shadow a text symbol that we do want? Eg. maybe in
> > one module we have
> > static int foo;
> > and in another we have
> > extern int foo(void);
> > Maybe they are even in separate objfiles (shared libs).
> > Maybe lookup_minimal_symbol will find the wrong one first,
> > and your change will terminate the search before we would
> > have found the right one?
> >
> > Just speculating...
>
> Well, we aren't using lookup_minimal_symbol; we're using
> lookup_minimal_symbol_by_pc_section. I don't think it's possible for
> it to get the wrong one, since it's by address instead of by name.
OK, good, that answers that concern. Thanks.
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [RFA] Fix for searching for data symbols
2002-01-19 9:33 [RFA] Fix for searching for data symbols Daniel Jacobowitz
2002-01-19 9:56 ` Michael Snyder
@ 2002-01-29 21:56 ` Daniel Jacobowitz
2002-01-30 15:26 ` Elena Zannoni
2 siblings, 0 replies; 7+ messages in thread
From: Daniel Jacobowitz @ 2002-01-29 21:56 UTC (permalink / raw)
To: gdb-patches
On Sat, Jan 19, 2002 at 12:33:18PM -0500, Daniel Jacobowitz wrote:
> This patch fixes the problem I described in:
> http://sources.redhat.com/ml/gdb/2002-01/msg00223.html
>
> Basically, since we use the bounds of the text segment to search for symbols
> in find_pc_*, don't let them return a bogus answer for a symbol we know is
> off in data land.
>
> OK to commit, or does anyone prefer a different approach to this?
Nobody ever commented on this, except for Michael Snyder's one
objection which I believe I've answered to his satisfaction. Could
this be approved, please?
> 2002-01-19 Daniel Jacobowitz <drow@mvista.com>
>
> * symtab.c (find_pc_sect_psymtab): Do not search psymtabs for
> data symbols, since we search based on textlow and texthigh.
> (find_pc_sect_symtab): Likewise.
>
> Index: symtab.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/symtab.c,v
> retrieving revision 1.51
> diff -u -p -r1.51 symtab.c
> --- symtab.c 2001/12/21 22:32:37 1.51
> +++ symtab.c 2002/01/19 17:29:09
> @@ -354,12 +354,22 @@ find_pc_sect_psymtab (CORE_ADDR pc, asec
> {
> register struct partial_symtab *pst;
> register struct objfile *objfile;
> + struct minimal_symbol *msymbol;
>
> + /* If we know that this is not a text address, return failure. */
> + msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> + if (msymbol
> + && (msymbol->type == mst_data
> + || msymbol->type == mst_bss
> + || msymbol->type == mst_abs
> + || msymbol->type == mst_file_data
> + || msymbol->type == mst_file_bss))
> + return NULL;
> +
> ALL_PSYMTABS (objfile, pst)
> {
> if (pc >= pst->textlow && pc < pst->texthigh)
> {
> - struct minimal_symbol *msymbol;
> struct partial_symtab *tpst;
>
> /* An objfile that has its functions reordered might have
> @@ -370,7 +380,6 @@ find_pc_sect_psymtab (CORE_ADDR pc, asec
> section == 0) /* can't validate section this way */
> return (pst);
>
> - msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> if (msymbol == NULL)
> return (pst);
>
> @@ -1385,6 +1394,17 @@ find_pc_sect_symtab (CORE_ADDR pc, asect
> register struct partial_symtab *ps;
> register struct objfile *objfile;
> CORE_ADDR distance = 0;
> + struct minimal_symbol *msymbol;
> +
> + /* If we know that this is not a text address, return failure. */
> + msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> + if (msymbol
> + && (msymbol->type == mst_data
> + || msymbol->type == mst_bss
> + || msymbol->type == mst_abs
> + || msymbol->type == mst_file_data
> + || msymbol->type == mst_file_bss))
> + return NULL;
>
> /* Search all symtabs for the one whose file contains our address, and which
> is the smallest of all the ones containing the address. This is designed
>
--
Daniel Jacobowitz Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [RFA] Fix for searching for data symbols
2002-01-19 9:33 [RFA] Fix for searching for data symbols Daniel Jacobowitz
2002-01-19 9:56 ` Michael Snyder
2002-01-29 21:56 ` Daniel Jacobowitz
@ 2002-01-30 15:26 ` Elena Zannoni
2002-01-30 21:04 ` Daniel Jacobowitz
2 siblings, 1 reply; 7+ messages in thread
From: Elena Zannoni @ 2002-01-30 15:26 UTC (permalink / raw)
To: Daniel Jacobowitz; +Cc: gdb-patches
Daniel Jacobowitz writes:
> This patch fixes the problem I described in:
> http://sources.redhat.com/ml/gdb/2002-01/msg00223.html
>
> Basically, since we use the bounds of the text segment to search for symbols
> in find_pc_*, don't let them return a bogus answer for a symbol we know is
> off in data land.
>
> OK to commit, or does anyone prefer a different approach to this?
>
I think it's fine. Just could you add comments with an explanation of
the odd case that you described in the mail above?
Thanks
Elena
> --
> Daniel Jacobowitz Carnegie Mellon University
> MontaVista Software Debian GNU/Linux Developer
>
> 2002-01-19 Daniel Jacobowitz <drow@mvista.com>
>
> * symtab.c (find_pc_sect_psymtab): Do not search psymtabs for
> data symbols, since we search based on textlow and texthigh.
> (find_pc_sect_symtab): Likewise.
>
> Index: symtab.c
> ===================================================================
> RCS file: /cvs/src/src/gdb/symtab.c,v
> retrieving revision 1.51
> diff -u -p -r1.51 symtab.c
> --- symtab.c 2001/12/21 22:32:37 1.51
> +++ symtab.c 2002/01/19 17:29:09
> @@ -354,12 +354,22 @@ find_pc_sect_psymtab (CORE_ADDR pc, asec
> {
> register struct partial_symtab *pst;
> register struct objfile *objfile;
> + struct minimal_symbol *msymbol;
>
> + /* If we know that this is not a text address, return failure. */
> + msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> + if (msymbol
> + && (msymbol->type == mst_data
> + || msymbol->type == mst_bss
> + || msymbol->type == mst_abs
> + || msymbol->type == mst_file_data
> + || msymbol->type == mst_file_bss))
> + return NULL;
> +
> ALL_PSYMTABS (objfile, pst)
> {
> if (pc >= pst->textlow && pc < pst->texthigh)
> {
> - struct minimal_symbol *msymbol;
> struct partial_symtab *tpst;
>
> /* An objfile that has its functions reordered might have
> @@ -370,7 +380,6 @@ find_pc_sect_psymtab (CORE_ADDR pc, asec
> section == 0) /* can't validate section this way */
> return (pst);
>
> - msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> if (msymbol == NULL)
> return (pst);
>
> @@ -1385,6 +1394,17 @@ find_pc_sect_symtab (CORE_ADDR pc, asect
> register struct partial_symtab *ps;
> register struct objfile *objfile;
> CORE_ADDR distance = 0;
> + struct minimal_symbol *msymbol;
> +
> + /* If we know that this is not a text address, return failure. */
> + msymbol = lookup_minimal_symbol_by_pc_section (pc, section);
> + if (msymbol
> + && (msymbol->type == mst_data
> + || msymbol->type == mst_bss
> + || msymbol->type == mst_abs
> + || msymbol->type == mst_file_data
> + || msymbol->type == mst_file_bss))
> + return NULL;
>
> /* Search all symtabs for the one whose file contains our address, and which
> is the smallest of all the ones containing the address. This is designed
^ permalink raw reply [flat|nested] 7+ messages in thread* Re: [RFA] Fix for searching for data symbols
2002-01-30 15:26 ` Elena Zannoni
@ 2002-01-30 21:04 ` Daniel Jacobowitz
0 siblings, 0 replies; 7+ messages in thread
From: Daniel Jacobowitz @ 2002-01-30 21:04 UTC (permalink / raw)
To: gdb-patches
On Wed, Jan 30, 2002 at 06:26:08PM -0500, Elena Zannoni wrote:
> Daniel Jacobowitz writes:
> > This patch fixes the problem I described in:
> > http://sources.redhat.com/ml/gdb/2002-01/msg00223.html
> >
> > Basically, since we use the bounds of the text segment to search for symbols
> > in find_pc_*, don't let them return a bogus answer for a symbol we know is
> > off in data land.
> >
> > OK to commit, or does anyone prefer a different approach to this?
> >
>
> I think it's fine. Just could you add comments with an explanation of
> the odd case that you described in the mail above?
Committed with comments, thanks!
--
Daniel Jacobowitz Carnegie Mellon University
MontaVista Software Debian GNU/Linux Developer
^ permalink raw reply [flat|nested] 7+ messages in thread