diff --git a/libexec/rtld-elf/map_object.c b/libexec/rtld-elf/map_object.c
index 333e08dbb943..f489eddd5610 100644
--- a/libexec/rtld-elf/map_object.c
+++ b/libexec/rtld-elf/map_object.c
@@ -61,7 +61,6 @@ map_object(int fd, const char *path, const struct stat *sb)
     Elf_Phdr **segs;
     int nsegs;
     Elf_Phdr *phdyn;
-    Elf_Phdr *phphdr;
     Elf_Phdr *phinterp;
     Elf_Phdr *phtls;
     caddr_t mapbase;
@@ -79,7 +78,8 @@ map_object(int fd, const char *path, const struct stat *sb)
     Elf_Addr clear_vaddr;
     caddr_t clear_addr;
     caddr_t clear_page;
-    size_t nclear;
+    Elf_Addr phdr_vaddr;
+    size_t nclear, phsize;
     Elf_Addr bss_vaddr;
     Elf_Addr bss_vlimit;
     caddr_t bss_addr;
@@ -95,9 +95,11 @@ map_object(int fd, const char *path, const struct stat *sb)
      * in that order.
      */
     phdr = (Elf_Phdr *) ((char *)hdr + hdr->e_phoff);
+    phsize  = hdr->e_phnum * sizeof (phdr[0]);
     phlimit = phdr + hdr->e_phnum;
     nsegs = -1;
-    phdyn = phphdr = phinterp = phtls = NULL;
+    phdyn = phinterp = phtls = NULL;
+    phdr_vaddr = 0;
     segs = alloca(sizeof(segs[0]) * hdr->e_phnum);
     while (phdr < phlimit) {
 	switch (phdr->p_type) {
@@ -108,7 +110,7 @@ map_object(int fd, const char *path, const struct stat *sb)
 
 	case PT_LOAD:
 	    segs[++nsegs] = phdr;
-    	    if (segs[nsegs]->p_align < PAGE_SIZE) {
+    	    if ((segs[nsegs]->p_align & (PAGE_SIZE - 1)) != 0) {
 		_rtld_error("%s: PT_LOAD segment %d not page-aligned",
 		    path, nsegs);
 		return NULL;
@@ -116,7 +118,8 @@ map_object(int fd, const char *path, const struct stat *sb)
 	    break;
 
 	case PT_PHDR:
-	    phphdr = phdr;
+	    phdr_vaddr = phdr->p_vaddr;
+	    phsize = phdr->p_memsz;
 	    break;
 
 	case PT_DYNAMIC:
@@ -211,6 +214,11 @@ map_object(int fd, const char *path, const struct stat *sb)
 		return NULL;
 	    }
 	}
+	if (phdr_vaddr == 0 && data_offset <= hdr->e_phoff &&
+	  (data_vlimit - data_vaddr + data_offset) >=
+	  (hdr->e_phoff + hdr->e_phnum * sizeof (Elf_Phdr))) {
+	    phdr_vaddr = data_vaddr + hdr->e_phoff - data_offset;
+	}
     }
 
     obj = obj_new();
@@ -227,10 +235,19 @@ map_object(int fd, const char *path, const struct stat *sb)
     obj->dynamic = (const Elf_Dyn *) (obj->relocbase + phdyn->p_vaddr);
     if (hdr->e_entry != 0)
 	obj->entry = (caddr_t) (obj->relocbase + hdr->e_entry);
-    if (phphdr != NULL) {
-	obj->phdr = (const Elf_Phdr *) (obj->relocbase + phphdr->p_vaddr);
-	obj->phsize = phphdr->p_memsz;
+    if (phdr_vaddr != 0) {
+	obj->phdr = (const Elf_Phdr *) (obj->relocbase + phdr_vaddr);
+    } else {
+	obj->phdr = malloc(phsize);
+	if (obj->phdr == NULL) {
+	    obj_free(obj);
+	    _rtld_error("%s: cannot allocate program header", path);
+	     return NULL;
+	}
+	memcpy((char *)obj->phdr, (char *)hdr + hdr->e_phoff, phsize);
+	obj->phdr_alloc = true;
     }
+    obj->phsize = phsize;
     if (phinterp != NULL)
 	obj->interp = (const char *) (obj->relocbase + phinterp->p_vaddr);
     if (phtls != NULL) {
@@ -307,7 +324,6 @@ obj_free(Obj_Entry *obj)
 
     if (obj->tls_done)
 	free_tls_offset(obj);
-    free(obj->path);
     while (obj->needed != NULL) {
 	Needed_Entry *needed = obj->needed;
 	obj->needed = needed->next;
@@ -328,9 +344,16 @@ obj_free(Obj_Entry *obj)
 	STAILQ_REMOVE_HEAD(&obj->dagmembers, link);
 	free(elm);
     }
-    free(obj->vertab);
-    free(obj->origin_path);
-    free(obj->priv);
+    if (obj->vertab)
+	free(obj->vertab);
+    if (obj->origin_path)
+	free(obj->origin_path);
+    if (obj->priv)
+	free(obj->priv);
+    if (obj->path)
+	free(obj->path);
+    if (obj->phdr_alloc)
+	free((void *)obj->phdr);
     free(obj);
 }
 
diff --git a/libexec/rtld-elf/rtld.c b/libexec/rtld-elf/rtld.c
index e8e31c579780..174c6ea3842d 100644
--- a/libexec/rtld-elf/rtld.c
+++ b/libexec/rtld-elf/rtld.c
@@ -164,6 +164,7 @@ static Obj_Entry **obj_tail;	/* Link field of last object in list */
 static Obj_Entry *obj_main;	/* The main program shared object */
 static Obj_Entry obj_rtld;	/* The dynamic linker shared object */
 static unsigned int obj_count;	/* Number of objects in obj_list */
+static unsigned int obj_loads;	/* Number of objects in obj_list */
 
 static Objlist list_global =	/* Objects dlopened with RTLD_GLOBAL */
   STAILQ_HEAD_INITIALIZER(list_global);
@@ -204,6 +205,7 @@ static func_ptr_type exports[] = {
     (func_ptr_type) &__tls_get_addr,
     (func_ptr_type) &_rtld_allocate_tls,
     (func_ptr_type) &_rtld_free_tls,
+    (func_ptr_type) &dl_iterate_phdr,
     NULL
 };
 
@@ -423,6 +425,7 @@ _rtld(Elf_Addr *sp, func_ptr_type *exit_proc, Obj_Entry **objp)
     *obj_tail = obj_main;
     obj_tail = &obj_main->next;
     obj_count++;
+    obj_loads++;
     /* Make sure we don't call the main program's init and fini functions. */
     obj_main->init = obj_main->fini = (Elf_Addr)NULL;
 
@@ -1387,6 +1390,7 @@ do_load_object(int fd, const char *name, char *path, struct stat *sbp)
     *obj_tail = obj;
     obj_tail = &obj->next;
     obj_count++;
+    obj_loads++;
     linkmap_add(obj);	/* for GDB & dlinfo() */
 
     dbg("  %p .. %p: %s", obj->mapbase,
@@ -2079,6 +2083,37 @@ dlinfo(void *handle, int request, void *p)
     return (error);
 }
 
+int
+dl_iterate_phdr(__dl_iterate_hdr_callback callback, void *param)
+{
+    struct dl_phdr_info phdr_info;
+    const Obj_Entry *obj;
+    int error, lockstate;
+
+    lockstate = rlock_acquire(rtld_bind_lock);
+
+    error = 0;
+
+    for (obj = obj_list;  obj != NULL;  obj = obj->next) {
+	phdr_info.dlpi_addr = (Elf_Addr)obj->relocbase;
+	phdr_info.dlpi_name = STAILQ_FIRST(&obj->names) ?
+	    STAILQ_FIRST(&obj->names)->name : obj->path;
+	phdr_info.dlpi_phdr = obj->phdr;
+	phdr_info.dlpi_phnum = obj->phsize / sizeof(obj->phdr[0]);
+	phdr_info.dlpi_tls_modid = obj->tlsindex;
+	phdr_info.dlpi_tls_data = obj->tlsinit;
+	phdr_info.dlpi_adds = obj_loads;
+	phdr_info.dlpi_subs = obj_loads - obj_count;
+
+	if ((error = callback(&phdr_info, sizeof phdr_info, param)) != 0)
+		break;
+
+    }
+    rlock_release(rtld_bind_lock, lockstate);
+
+    return (error);
+}
+
 struct fill_search_info_args {
     int		 request;
     unsigned int flags;
diff --git a/libexec/rtld-elf/rtld.h b/libexec/rtld-elf/rtld.h
index cc430e86f8f4..c7be5235578a 100644
--- a/libexec/rtld-elf/rtld.h
+++ b/libexec/rtld-elf/rtld.h
@@ -201,15 +201,16 @@ typedef struct Struct_Obj_Entry {
     Elf_Addr init;		/* Initialization function to call */
     Elf_Addr fini;		/* Termination function to call */
 
-    bool mainprog;		/* True if this is the main program */
-    bool rtld;			/* True if this is the dynamic linker */
-    bool textrel;		/* True if there are relocations to text seg */
-    bool symbolic;		/* True if generated with "-Bsymbolic" */
-    bool bind_now;		/* True if all relocations should be made first */
-    bool traced;		/* Already printed in ldd trace output */
-    bool jmpslots_done;		/* Already have relocated the jump slots */
-    bool init_done;		/* Already have added object to init list */
-    bool tls_done;		/* Already allocated offset for static TLS */
+    bool mainprog : 1;		/* True if this is the main program */
+    bool rtld : 1;		/* True if this is the dynamic linker */
+    bool textrel : 1;		/* True if there are relocations to text seg */
+    bool symbolic : 1;		/* True if generated with "-Bsymbolic" */
+    bool bind_now : 1;		/* True if all relocations should be made first */
+    bool traced : 1;		/* Already printed in ldd trace output */
+    bool jmpslots_done : 1;	/* Already have relocated the jump slots */
+    bool init_done : 1;		/* Already have added object to init list */
+    bool tls_done : 1;		/* Already allocated offset for static TLS */
+    bool phdr_alloc : 1;	/* Phdr is allocated and needs to be freed. */
 
     struct link_map linkmap;	/* for GDB and dlinfo() */
     Objlist dldags;		/* Object belongs to these dlopened DAGs (%) */