rtld_next broke?

Timo Teras timo.teras at iki.fi
Fri Nov 2 06:52:39 UTC 2012


On Thu, 1 Nov 2012 21:13:32 +0200 Timo Teras <timo.teras at iki.fi> wrote:
> I noticed that dlsym(RTLD_NEXT) seems to be broken under certain
> conditions. I also noticed that Floarian observed this earlier too
> [1], but no solution seems to have been provided.
> 
> It appears this is related closely to the linking order of the final
> executable.
>[snip]
> 
> Appears that we need to add some sort of for loop iterating through
> all the following modules.
> 
> Will try to patch tomorrow. Or does someone have a better solution for
> this?

I'm now playing with the following patch. Fixes my trivial test case.

--- a/ldso/libdl/libdl.c
+++ b/ldso/libdl/libdl.c
@@ -671,7 +671,7 @@
 {
 	struct elf_resolve *tpnt, *tfrom;
 	struct dyn_elf *handle;
-	ElfW(Addr) from;
+	ElfW(Addr) from = 0;
 	struct dyn_elf *rpnt;
 	void *ret;
 	struct symbol_ref sym_ref = { NULL, NULL };
@@ -729,7 +729,13 @@
 	tpnt = NULL;
 	if (handle == _dl_symbol_tables)
 		tpnt = handle->dyn; /* Only search RTLD_GLOBAL objs if global object */
-	ret = _dl_find_hash(name2, &handle->dyn->symbol_scope, tpnt, ELF_RTYPE_CLASS_DLSYM, &sym_ref);
+
+	do {
+		ret = _dl_find_hash(name2, &handle->dyn->symbol_scope, tpnt, ELF_RTYPE_CLASS_DLSYM, &sym_ref);
+		if (ret != NULL)
+			break;
+		handle = handle->next;
+	} while (from && handle);
 
 #if defined(USE_TLS) && USE_TLS && defined SHARED
 	if (sym_ref.sym && (ELF_ST_TYPE(sym_ref.sym->st_info) == STT_TLS) && (sym_ref.tpnt)) {



More information about the uClibc mailing list