Mirror of the gdb mailing list
 help / color / mirror / Atom feed
From: Andrew Burgess <andrew.burgess@embecosm.com>
To: Mathieu Tarral <mathieu.tarral@protonmail.com>
Cc: "gdb@sourceware.org" <gdb@sourceware.org>
Subject: Re: Force software breakpoint on read-only memory
Date: Wed, 21 Nov 2018 15:02:00 -0000	[thread overview]
Message-ID: <20181121150238.GZ16539@embecosm.com> (raw)
In-Reply-To: <UpeoDeMdWlsVJSDXnbjmxHp8zp3Qx_idZKmlMb5AElmjUCmP7cMVlZJSRhhgpoKUqTPFSVGBAuiyYnlu8PkQnA_1cjC6gGwELiT5pQqXPGY=@protonmail.com>

* Mathieu Tarral <mathieu.tarral@protonmail.com> [2018-11-21 09:40:15 +0000]:

> Hi,
> 
> I'm trying to set a software breakpoint on a read-only memory page.
> GDB automatically sets the breakpoint as hardware, but I would like to
> keep using software breakpoints.
> 
> I found the `set breakpoint auto-hw off option in` in the
> documentation:
> "If the target provides a memory map, GDB will warn when
> trying to set software breakpoint at a read-only address."
> 
> However, even when I set it, GDB gives me a warning and my command is aborted:
> 
> $ gdb -x gdbinit ~/tmp/ntoskrnl.exe
> 0x80545c9f in ?? ()
> (gdb) set breakpoint auto-hw off
> (gdb) b NtOpenFile
> Breakpoint 1 at 0x4a30b1
> (gdb) continue
> Continuing.
> Warning:
> Cannot insert breakpoint 1.
> Cannot set software breakpoint at read-only address 0x4a30b1
> 
> Command aborted.
> (gdb)
> 
> 
> 
> My stub interacts directly with the guest hardware, from the
> hypervisor, so I can bypass the read-only protections.
> That's why i wanted to force the software breakpoint
> and ignore the memory map.
> 
> Is this a bug in GDB ?

Not in the sense that this is the designed behaviour.  Once GDB has a
memory map it assumes this is accurate.  When it spots you trying to
place a breakpoint in read-only memory if checks the 'auto-hw'
setting.  If this is on then it converts to a h/w breakpoint,
otherwise, it fails, after all you can't write to read-only memory.

The place to look in the GDB source is gdb/breakpoint.c:insert_bp_location.

My question would be, do you allow _all_ writes to read-only memory,
or only breakpoint writes?  If it's all, then could you "fix" the
memory map?

If you only want to allow s/w breakpoint writes then you'll need to
hack GDB, possibly adding a switch to ignore memory maps for s/w
breakpoints.

Thanks,
Andrew


> 
> Thanks !
> --
> Mathieu Tarral
> 
> Sent with ProtonMail Secure Email.


      reply	other threads:[~2018-11-21 15:02 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-11-21  9:40 Mathieu Tarral
2018-11-21 15:02 ` Andrew Burgess [this message]

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=20181121150238.GZ16539@embecosm.com \
    --to=andrew.burgess@embecosm.com \
    --cc=gdb@sourceware.org \
    --cc=mathieu.tarral@protonmail.com \
    /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