Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
From: Jan Kratochvil <jan.kratochvil@redhat.com>
To: gdb-patches@sourceware.org
Subject: [patch] infcall: Remove gdb_assert ($sp overflow)
Date: Fri, 19 Feb 2010 22:48:00 -0000	[thread overview]
Message-ID: <20100219224805.GA9681@host0.dyn.jankratochvil.net> (raw)

Hi,

set $sp=0
call something()
->
../../gdb/infcall.c:521: internal-error: call_function_by_hand: Assertion
`(gdbarch_inner_than (gdbarch, 1, 2) && sp <= old_sp) || (gdbarch_inner_than
(gdbarch, 2, 1) && sp >= old_sp)' failed.

as $sp - frame == 0xffffsmth which is not lower than $sp.

It must not be gdb_assert().  It can be an error() but I left it just to do:
	(gdb) set $sp=0
	(gdb) call doubleit (1)
	Cannot access memory at address 0xffffffffffffff78
	(gdb) set $sp=-1
	(gdb) call doubleit (1)
	Cannot access memory at address 0xffffffffffffff68

Unaware how inconvenient is an intentional crash of the testcase on some
embedded/non-MMU systems.
No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.


Thanks,
Jan


gdb/
2010-02-19  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* infcall.c (call_function_by_hand): Remove gdb_assert on sp and old_sp.
	New comment.

gdb/testsuite/
2010-02-19  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* gdb.base/callfuncs.exp: New tests for $spval 0 and -1.  Remove return.

--- a/gdb/infcall.c
+++ b/gdb/infcall.c
@@ -518,10 +518,9 @@ call_function_by_hand (struct value *function, int nargs, struct value **args)
 	      /* Stack grows up.  */
 	      sp = gdbarch_frame_align (gdbarch, old_sp + 1);
 	  }
-	gdb_assert ((gdbarch_inner_than (gdbarch, 1, 2)
-		    && sp <= old_sp)
-		    || (gdbarch_inner_than (gdbarch, 2, 1)
-		       && sp >= old_sp));
+	/* SP may have overflown address zero here from OLD_SP.  Memory access
+	   functions will probably fail in such case but that is a target's
+	   problem.  */
       }
     else
       /* FIXME: cagney/2002-09-18: Hey, you loose!
--- a/gdb/testsuite/gdb.base/callfuncs.exp
+++ b/gdb/testsuite/gdb.base/callfuncs.exp
@@ -469,5 +469,18 @@ if {$old_reg_content == $new_reg_content} then {
     fail "nested call dummies preserve register contents"
 }
 
-return 0
+# GDB should not crash by internal error on $sp overflow during the inferior
+# call.  It is OK it will stop on some: Cannot access memory at address 0x$hex.
 
+foreach spval {0 -1} {
+    set old_ldprefix $pf_prefix
+    lappend pf_prefix "sp=$spval:"
+
+    gdb_test {set $old_sp = $sp}
+    gdb_test "set \$sp = $spval"
+
+    gdb_test "call doubleit (1)"
+
+    gdb_test {set $sp = $old_sp}
+    set pf_prefix $old_ldprefix
+}


             reply	other threads:[~2010-02-19 22:48 UTC|newest]

Thread overview: 8+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-02-19 22:48 Jan Kratochvil [this message]
2010-02-26 22:45 ` Tom Tromey
2010-02-27  1:19   ` [patch] infcall: Remove gdb_assert ($sp underflow) Jan Kratochvil
2010-02-28 10:53   ` [patch] infcall: Remove gdb_assert ($sp overflow) Joel Brobecker
2010-02-26 22:53 ` Daniel Jacobowitz
2010-02-27  0:46   ` [patch] infcall: Remove gdb_assert ($sp underflow) Jan Kratochvil
2010-02-28 14:35     ` Daniel Jacobowitz
2010-02-28 17:58       ` Jan Kratochvil

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=20100219224805.GA9681@host0.dyn.jankratochvil.net \
    --to=jan.kratochvil@redhat.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