This is the mail archive of the
gdb-prs@sourceware.org
mailing list for the GDB project.
[Bug server/14922] New: gdbserver: single-stepping out of signaltrampolines on Linux/i386
- From: "palves at redhat dot com" <sourceware-bugzilla at sourceware dot org>
- To: gdb-prs at sourceware dot org
- Date: Thu, 06 Dec 2012 19:48:31 +0000
- Subject: [Bug server/14922] New: gdbserver: single-stepping out of signaltrampolines on Linux/i386
- Auto-submitted: auto-generated
http://sourceware.org/bugzilla/show_bug.cgi?id=14922
Bug #: 14922
Summary: gdbserver: single-stepping out of signal trampolines
on Linux/i386
Product: gdb
Version: HEAD
Status: NEW
Severity: normal
Priority: P2
Component: server
AssignedTo: unassigned@sourceware.org
ReportedBy: palves@redhat.com
Classification: Unclassified
Host: i686-pc-linux-gnu
Target: i686-pc-linux-gnu
i386-linux-nat.c:i386_linux_resume has special handling for stepping out of a
signal handler, introduced by
http://sourceware.org/ml/gdb-patches/2000-04/msg00582.html.
That code is used when native debugging.
However, GDBserver doesn't have this handling, but, stepping out of the signal
trampoline seems to work correctly anyway.
For testing, I made sure I got to the signal handler with a normal continue, so
that the restored context would possibly not have the trace flag set. I
confirmed the i386-linux-nat.c code is still reachable (Fedora 17, x86_64
-m32).
Perhaps the kernel is already taking care of leaving the trace flag set
nowadays, if it sees it set on a sigreturn/rt_sigreturn entry.
This is relevant for the local/remote feature parity project.
--
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.