[PATCH][PING][PR gdb/19374] null-ptr dereference on solaris when running get_osdata( "processes" )

[Brian Vandenberg <phantall@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1030 bytes --]

This patch is to address bug 19374 on solaris.

When running gdb under eclipse/CDT on Solaris, eclipse runs gdb with:

gdb --interpreter mi2 --nx

During execution it ends up sending the following string to gdb:

16-list-thread-groups --available

... at which point gdb crashes.

This patch partially reverts the change in the following:

http://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commitdiff;h=e75fdfcad1c868eae5396a95be9dd18010406306#patch4

... by re-adding the check for ops->beneath != NULL and returning
TARGET_XFER_E_IO.

I cannot easily run the test suite against these changes right now.
If this gets rejected based on that, I'll see about getting something
setup at home to run the tests when I have time.

----

note: this patch was tested against 8.1.1.  I don't know [yet] whether
this is still a problem in 8.2 because of the procfs C++ conversion.
The same logic still exists but "beneath" appears to be a function
object.

-brian

ps, my assignment/release forms were completed/received 10/30/2017

[-- Attachment #2: gdb-19374-patch, revised.txt --]
[-- Type: text/plain, Size: 742 bytes --]

gdb/Changelog:
2018-08-07  Brian Vandenberg <phantall@gmail.com>

	PR gdb/19374
	* gdb/procfs.c (procfs_xfer_partial): Added check for ops->beneath != NULL

diff --git a/gdb/procfs.c b/gdb/procfs.c
--- a/gdb/procfs.c
+++ b/gdb/procfs.c
@@ --2599,9 +2599,12 @@ procfs_xfer_partial (struct target_ops *ops, enum target_object object,
 #endif

     default:
-      return ops->beneath->to_xfer_partial (ops->beneath, object, annex,
+      if( ops->beneath )
+        return ops->beneath->to_xfer_partial (ops->beneath, object, annex,
                                            readbuf, writebuf, offset, len,
                                            xfered_len);
+      else
+        return TARGET_XFER_E_IO;
     }
 }