Source for 3.2.0-3

[Byron Hawkins]

I'm trying to build DynamoRIO version 3.2.0-3, which is the latest binary distribution. But the tags in svn only mark up to the minor version numbers, so the closest I can get is release_3_2_0, which has build errors and a deadlock (per issue 762). I'd prefer to have the latest stable build, rather than trying to work from the head. Can someone tell me how I can get the source that 3.2.0-3 was actually built from? Thanks for your help.

Byron

[Derek Bruening]

It was built from the release_3_2_0 tag, and it presumably has the same deadlock with that version of glibc.  It was built with an older toolchain so it did not hit the Ubuntu 12.04 build errors.  The head is pretty stable and has bug fixes not in the 3.2.0 release.  Only if you start using new API routines that are in flux between releases will there be an issue.

On Thu, Aug 30, 2012 at 2:39 PM, Byron Hawkins <by...@hawkinssoftware.net> wrote:

I'm trying to build DynamoRIO version 3.2.0-3, which is the latest binary distribution. But the tags in svn only mark up to the minor version numbers, so the closest I can get is release_3_2_0, which has build errors and a deadlock (per issue 762). I'd prefer to have the latest stable build, rather than trying to work from the head. Can someone tell me how I can get the source that 3.2.0-3 was actually built from? Thanks for your help.

Byron

--
You received this message because you are subscribed to the Google Groups "DynamoRIO Users" group.
To view this discussion on the web visit https://groups.google.com/d/msg/dynamorio-users/-/4YotQB18M0sJ.
To post to this group, send email to dynamor...@googlegroups.com.
To unsubscribe from this group, send email to dynamorio-use...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/dynamorio-users?hl=en.

[Byron Hawkins]

Hi Derek, thanks for your reply. The strange thing is that I get the deadlock consistently with my build from source, but never with the binary distribution. I run them both on the same Ubuntu 12.04 machine, and simply switching $DYNAMORIO_HOME effectively toggles the deadlock on and off. It makes me wonder, can you tell me what the "-3" means in "3.2.0-3"?  

For now I will try building the latest code. Thanks for letting me know what to expect, we are only using ordinary features of DynamoRIO at this point, so the latest commits should work fine for us.

Byron

[Derek Bruening]

On Thu, Aug 30, 2012 at 10:56 PM, Byron Hawkins <by...@hawkinssoftware.net> wrote:

Hi Derek, thanks for your reply. The strange thing is that I get the deadlock consistently with my build from source, but never with the binary distribution.

The binary was built from an older toolchain explicitly to use older versioned glibc imports which may explain this.

 

I run them both on the same Ubuntu 12.04 machine, and simply switching $DYNAMORIO_HOME effectively toggles the deadlock on and off. It makes me wonder, can you tell me what the "-3" means in "3.2.0-3"?  

Just the release candidate build iteration.  -1 and -2 were prior svn versions that didn't make the cut.

 

For now I will try building the latest code. Thanks for letting me know what to expect, we are only using ordinary features of DynamoRIO at this point, so the latest commits should work fine for us.

You can check the release.dox for any target version for compatibility changes.  We try to avoid breaking back compat at all costs and we update release.dox with API-visible changes.

- Derek

[James Spooner]

Hi Derek,

For the record, which version of GCC/GLIBC did you build 3.2.0-3 with?  I'm seeing segfaults in privload_tls_init() because dr_tp is being shifted 16 bytes from the end of the page, so dr_tp->self crosses the boundary (I note that this code seems to be re-worked as of r1569).  Using 3.2.0-3 its fine, this is a CentOS 5.5 box, gcc version 4.1.2 20080704 (Red Hat 4.1.2-48), glibc-2.5-49.el5_5.4.   

Additionally, I need to comment out the reference to 'xgetbv' in signal.c, (gcc says it's an invalid instruction unless I compile with gcc 4.7.0).    And at HEAD (r1569) I can't resolve the reference to AT_EXECFN, presumably this is new?

Cheers,

James

--
You received this message because you are subscribed to the Google Groups "DynamoRIO Users" group.

To post to this group, send email to dynamor...@googlegroups.com.
To unsubscribe from this group, send email to dynamorio-use...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/dynamorio-users?hl=en.

--
James Spooner
+44 (7842) 772 423