This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: rfc patch for buildid < shlib base address


On Sun, 2009-03-15 at 23:54 +0100, Mark Wielaard wrote:
> And digging deeper I see roland already fixed this issue a couple of
> days ago: https://bugzilla.redhat.com/show_bug.cgi?id=489439
> 
> I'll look into adding the workaround described in that bug to our code
> tomorrow.

Added as:

commit 1bb61ae1fd02086560f5cd019db320b5a217ba05
Author: Mark Wielaard <mjw@redhat.com>
Date:   Mon Mar 16 14:19:20 2009 +0100

Add workaround for dwfl_module_build_id bug with elfutils < 0.140.
    
* translate.cxx (dump_unwindsyms): Check elfutils version and whether
  build_id_vaddr < base, and if so add main_bias to address.

Note that it checks both for !_ELFUTILS_PREREQ(0,141) and for
build_id_vaddr < base because it looks like the buggy behavior doesn't
occur always with older elfutils versions and if it doesn't occur, then
adding main.bias is wrong. I assume that if the bug occurs then
build_id_vaddr will always be smaller than main.bias since it is
supposed to go as close as possible to the start of the elf file.

Cheers,

Mark
diff --git a/translate.cxx b/translate.cxx
index 17c37dc..f4c2853 100644
--- a/translate.cxx
+++ b/translate.cxx
@@ -4497,17 +4497,35 @@ dump_unwindsyms (Dwfl_Module *m,
     // see https://bugzilla.redhat.com/show_bug.cgi?id=465872
     // and http://sourceware.org/ml/systemtap/2008-q4/msg00579.html
 #ifdef _ELFUTILS_PREREQ
-#if _ELFUTILS_PREREQ(0,138)
+  #if _ELFUTILS_PREREQ(0,138)
     // Let's standardize to the buggy "end of build-id bits" behavior. 
     build_id_vaddr += build_id_len;
+  #endif
+  #if !_ELFUTILS_PREREQ(0,141)
+    #define NEED_ELFUTILS_BUILDID_WORKAROUND
+  #endif
+#else
+  #define NEED_ELFUTILS_BUILDID_WORKAROUND
 #endif
+
+    // And check for another workaround needed.
+    // see https://bugzilla.redhat.com/show_bug.cgi?id=489439
+    // and http://sourceware.org/ml/systemtap/2009-q1/msg00513.html
+#ifdef NEED_ELFUTILS_BUILDID_WORKAROUND
+    if (build_id_vaddr < base && dwfl_module_relocations (m) == 1)
+      {
+        GElf_Addr main_bias;
+        dwfl_module_getelf (m, &main_bias);
+        build_id_vaddr += main_bias;
+      }
 #endif
-        if (c->session.verbose > 1) {
-           clog << "Found build-id in " << name
-                << ", length " << build_id_len;
-           clog << ", end at 0x" << hex << build_id_vaddr
-                << dec << endl;
-        }
+    if (c->session.verbose > 1)
+      {
+        clog << "Found build-id in " << name
+             << ", length " << build_id_len;
+        clog << ", end at 0x" << hex << build_id_vaddr
+             << dec << endl;
+      }
   }
 
   // Look up the relocation basis for symbols

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]