From: Joel Brobecker <brobecker@adacore.com>
To: cgf
Cc: gdb-patches@sourceware.org
Subject: [WINDOWS/RFC] Invalid segment resister value on x86_64-windows
Date: Wed, 02 May 2012 21:58:00 -0000 [thread overview]
Message-ID: <20120502215738.GE15555@adacore.com> (raw)
In-Reply-To: <20120502212726.GA24839@adacore.com>
Hi Chris,
Would you mind telling us what you think on this issue?
Thanks!
> > > Hmm, it should be possible to do this without changing any of the
> > > -tdep.c code, but if you prefer to do it this way, can you rename the
> > > new field into cs_regnum, change the comment into "Register number for
> > > %cs", and move it somewhere around the other _regnum variables that
> > > are already part of the struct? That makes things a bit more
> > > consistent.
> >
> > If you prefer, I can do something similar to what we do to handle
> > the register mapping. Attached is a patch that does that, and only
> > touches windows *-nat code. Untested for now; will test and add
> > a ChangeLog if this is your prefered approach.
> >
> > This can be simplified a little further and have everything done
> > in windows-nat.c, but it would probably be considered uglier:
> > include "i386-tdep.h" and "amd64-tdep.h" in windows-nat.c, and
> > then have a quick check to determine which platform we are, and
> > then set then implement the segment_register_p function using either
> > a check on sizeof (void *), or a #ifdef [...] #else [...]...
commit ca26ab21b1b2d080e9002d86f23e3e41161af6de
Author: Joel Brobecker <brobecker@adacore.com>
Date: Wed May 2 14:19:23 2012 -0700
Segment register reading on Windows targets.
diff --git a/gdb/amd64-windows-nat.c b/gdb/amd64-windows-nat.c
index bc2c047..e5fb0e0 100644
--- a/gdb/amd64-windows-nat.c
+++ b/gdb/amd64-windows-nat.c
@@ -18,6 +18,8 @@
#include "defs.h"
#include "windows-nat.h"
#include "i386-nat.h"
+#include "amd64-tdep.h"
+
#include <windows.h>
#define context_offset(x) (offsetof (CONTEXT, x))
@@ -85,6 +87,14 @@ static const int mappings[] =
};
#undef context_offset
+/* segment_register_p_ftype implementation for amd64. */
+
+static int
+amd64_windows_segment_register_p (int regnum)
+{
+ return regnum >= AMD64_CS_REGNUM && regnum <= AMD64_GS_REGNUM;
+}
+
/* -Wmissing-prototypes */
extern initialize_file_ftype _initialize_amd64_windows_nat;
@@ -92,5 +102,6 @@ void
_initialize_amd64_windows_nat (void)
{
windows_set_context_register_offsets (mappings);
+ windows_set_segment_register_p (amd64_windows_segment_register_p);
i386_set_debug_register_length (8);
}
diff --git a/gdb/i386-windows-nat.c b/gdb/i386-windows-nat.c
index 5d93915..0928c9f 100644
--- a/gdb/i386-windows-nat.c
+++ b/gdb/i386-windows-nat.c
@@ -18,6 +18,7 @@
#include "defs.h"
#include "windows-nat.h"
#include "i386-nat.h"
+#include "i386-tdep.h"
#include <windows.h>
@@ -70,6 +71,14 @@ static const int mappings[] =
};
#undef context_offset
+/* segment_register_p_ftype implementation for x86. */
+
+static int
+i386_windows_segment_register_p (int regnum)
+{
+ return regnum >= I386_CS_REGNUM && regnum <= I386_GS_REGNUM;
+}
+
/* -Wmissing-prototypes */
extern initialize_file_ftype _initialize_i386_windows_nat;
@@ -77,5 +86,6 @@ void
_initialize_i386_windows_nat (void)
{
windows_set_context_register_offsets (mappings);
+ windows_set_segment_register_p (i386_windows_segment_register_p);
i386_set_debug_register_length (4);
}
diff --git a/gdb/windows-nat.c b/gdb/windows-nat.c
index f536ed1..000c86f 100644
--- a/gdb/windows-nat.c
+++ b/gdb/windows-nat.c
@@ -243,6 +243,10 @@ static int useshell = 0; /* use shell for subprocesses */
static const int *mappings;
+/* The function to use in order to determine whether a register is
+ a segment register or not. */
+static segment_register_p_ftype *segment_register_p;
+
/* This vector maps the target's idea of an exception (extracted
from the DEBUG_EVENT structure) to GDB's idea. */
@@ -272,6 +276,14 @@ windows_set_context_register_offsets (const int *offsets)
mappings = offsets;
}
+/* See windows-nat.h. */
+
+void
+windows_set_segment_register_p (segment_register_p_ftype *fun)
+{
+ segment_register_p = fun;
+}
+
static void
check (BOOL ok, const char *file, int line)
{
@@ -456,6 +468,14 @@ do_windows_fetch_inferior_registers (struct regcache *regcache, int r)
l = (*((long *) context_offset) >> 16) & ((1 << 11) - 1);
regcache_raw_supply (regcache, r, (char *) &l);
}
+ else if (segment_register_p (r))
+ {
+ /* GDB treats segment registers as 32bit registers, but they are
+ in fact only 16 bits long. Make sure we do not read extra
+ bits from our source buffer. */
+ l = *((long *) context_offset) & 0xffff;
+ regcache_raw_supply (regcache, r, (char *) &l);
+ }
else if (r >= 0)
regcache_raw_supply (regcache, r, context_offset);
else
diff --git a/gdb/windows-nat.h b/gdb/windows-nat.h
index 08200b9..a6cc5ec 100644
--- a/gdb/windows-nat.h
+++ b/gdb/windows-nat.h
@@ -20,5 +20,13 @@
extern void windows_set_context_register_offsets (const int *offsets);
+/* A pointer to a function that should return non-zero iff REGNUM
+ corresponds to one of the segment registers. */
+typedef int (segment_register_p_ftype) (int regnum);
+
+/* Set the function that should be used by this module to determine
+ whether a given register is a segment register or not. */
+extern void windows_set_segment_register_p (segment_register_p_ftype *fun);
+
#endif
--
Joel
next prev parent reply other threads:[~2012-05-02 21:58 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-05-01 23:04 Joel Brobecker
2012-05-01 23:04 ` [RFA/commit 1/2] Regenerate the features/i386 target description .c files Joel Brobecker
2012-05-02 6:34 ` Sergio Durigan Junior
2012-05-01 23:05 ` [RFA 2/2] [x86/x86_64] Segment registers are 16 bits long (not 32bits) Joel Brobecker
2012-05-02 0:01 ` [RFA 3/2(+)] Test size of x86/x86_64 segment registers Joel Brobecker
2012-05-02 10:10 ` Invalid segment resister value on x86_64-windows Mark Kettenis
2012-05-02 17:57 ` Joel Brobecker
2012-05-02 20:45 ` Mark Kettenis
2012-05-02 21:26 ` Joel Brobecker
2012-05-02 21:27 ` Joel Brobecker
2012-05-02 21:50 ` Mark Kettenis
2012-05-02 21:58 ` Joel Brobecker [this message]
2012-05-02 22:10 ` [WINDOWS/RFC] " Christopher Faylor
2012-05-02 22:16 ` Christopher Faylor
2012-05-04 18:38 ` checked in: " Joel Brobecker
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20120502215738.GE15555@adacore.com \
--to=brobecker@adacore.com \
--cc=gdb-patches@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox