Re: Merging scratch/no-purespace to remove unexec and purespace

[Pip Cet]

"Andrea Corallo" <acorallo@gnu.org> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>>> Date: Sat, 04 Jan 2025 11:12:48 +0000
>>> From: Pip Cet <pipcet@protonmail.com>
>>> Cc: emacs-devel@gnu.org, stefankangas@gmail.com, acorallo@gnu.org, 
>>> gerd.moellmann@gmail.com, monnier@iro.umontreal.ca
>>>
>>> "Eli Zaretskii" <eliz@gnu.org> writes:
>>>
>>> > Is this about bug#74966?  If so, let's not discuss the same issue
>>>
>>> No.  This is about the --enable-checking code treating --lambda-fixup as
>>> a forbidden symbol and crashing if it is used in the obvious fashion by
>>> any code that happens to be native-compiled, which is the current status
>>> on scratch/no-purespace.
>>
>> Then I'd like to hear Andrea's opinion on this.
>
> My opinion is that, to fix the scratch/no-purespace bug, I'm okay with
> the #$ solution if it works.  Honestly I'd be fine also using a comp
> specific symbol like comp--something which is documented not to be
> compilable, maybe less nice but probably less convoluted.  Anyway if Pip
> prefers #$ and is not too complex I'm okay with that.

Here's the patch.  Okay to push?

(As with all the DEFVARs in comp.c, space for the new symbol is reserved
in globals, and the documentation ends up in etc/DOC, even in
non-nativecomp builds.  The same is true for DEFSYM: make-docfile.c
doesn't care about #ifdef.  If someone fixes this so non-nativecomp
builds don't pay a price for nativecomp, this code will need fixing,
too.)

commit a0abd3033414a011125e815d30890ed22b80a84e (HEAD)
Author: Pip Cet <pipcet@protonmail.com>
Date:   Tue Jan 7 12:13:40 2025 +0000

    Use #$ for lambda fixups in native compilation data vectors
    
    * lisp/emacs-lisp/comp.el (comp--#$): New defvar.
    (comp--finalize-container): Use it.
    * src/comp.c (ABI_VERSION): Bump.
    (emit_static_object): Ensure 'comp--#$' prints as "#$".
    (load_static_obj): Ensure '#$' reads as Vcomp__hashdollar.
    (check_comp_unit_relocs): Adjust assertion.
    (syms_of_comp): Define 'comp--#$'.
    * src/pdumper.c (dump_do_dump_relocation): Adjust assertion.

diff --git a/lisp/emacs-lisp/comp.el b/lisp/emacs-lisp/comp.el
index ab6fd77f11a..a8698ef6454 100644
--- a/lisp/emacs-lisp/comp.el
+++ b/lisp/emacs-lisp/comp.el
@@ -42,6 +42,7 @@ comp-native-version-dir
 (defvar comp-subr-arities-h)
 (defvar native-comp-eln-load-path)
 (defvar native-comp-enable-subr-trampolines)
+(defvar comp--\#$)
 
 (declare-function comp--compile-ctxt-to-file0 "comp.c")
 (declare-function comp--init-ctxt "comp.c")
@@ -3254,10 +3255,9 @@ comp--finalize-container
                  ;; from the corresponding m-var.
                  collect (if (gethash obj
                                       (comp-ctxt-byte-func-to-func-h 
comp-ctxt))
-                             ;; Hack not to have `--lambda-fixup' in
-                             ;; data relocations as it would trigger the
-                             ;; check in 'check_comp_unit_relocs'.
-                             (intern (concat (make-string 1 ?-) 
"-lambda-fixup"))
+                             ;; This prints as #$, so we can assert this
+                             ;; value does not remain in the data vector
+                             comp--\#$
                            obj))))
 
 (defun comp--finalize-relocs ()
diff --git a/src/comp.c b/src/comp.c
index 8b38adec252..ecdf01b91cb 100644
--- a/src/comp.c
+++ b/src/comp.c
@@ -468,7 +468,7 @@ load_gccjit_if_necessary (bool mandatory)
 
 
 /* Increase this number to force a new Vcomp_abi_hash to be generated.  */
-#define ABI_VERSION "9"
+#define ABI_VERSION "10"
 
 /* Length of the hashes used for eln file naming.  */
 #define HASH_LENGTH 8
@@ -2682,6 +2682,12 @@ emit_static_object (const char *name, Lisp_Object obj)
   specbind (intern_c_string ("print-quoted"), Qt);
   specbind (intern_c_string ("print-gensym"), Qt);
   specbind (intern_c_string ("print-circle"), Qt);
+  /* Bind print-number-table and print-continuous-numbering so comp--#$
+     prints as #$.  */
+  Lisp_Object print_number_table = CALLN (Fmake_hash_table, QCtest, Qeq);
+  Fputhash (Vcomp__hashdollar, build_string ("#$") , print_number_table);
+  specbind (intern_c_string ("print-number-table"), print_number_table);
+  specbind (intern_c_string ("print-continuous-numbering"), Qt);
   Lisp_Object str = Fprin1_to_string (obj, Qnil, Qnil);
   unbind_to (count, Qnil);
 
@@ -5145,18 +5151,22 @@ fixup_eln_load_path (Lisp_Object eln_filename)
 static Lisp_Object
 load_static_obj (struct Lisp_Native_Comp_Unit *comp_u, const char *name)
 {
+  specpdl_ref count = SPECPDL_INDEX ();
   static_obj_t *blob =
     dynlib_sym (comp_u->handle, format_string ("%s_blob", name));
+  /* Special value so we can recognize #$.  */
+  specbind (intern_c_string ("load-file-name"),
+           Vcomp__hashdollar);
   if (blob)
     /* New blob format.  */
-    return Fread (make_string (blob->data, blob->len));
+    return unbind_to (count, Fread (make_string (blob->data, blob->len)));
 
   static_obj_t *(*f)(void) = dynlib_sym (comp_u->handle, name);
   if (!f)
     xsignal1 (Qnative_lisp_file_inconsistent, comp_u->file);
 
   blob = f ();
-  return Fread (make_string (blob->data, blob->len));
+  return unbind_to (count, Fread (make_string (blob->data, blob->len)));
 
 }
 
@@ -5173,7 +5183,7 @@ check_comp_unit_relocs (struct Lisp_Native_Comp_Unit 
*comp_u)
   for (ptrdiff_t i = 0; i < d_vec_len; i++)
     {
       Lisp_Object x = data_relocs[i];
-      if (EQ (x, Q__lambda_fixup))
+      if (EQ (x, Vcomp__hashdollar))
        return false;
       else if (NATIVE_COMP_FUNCTIONP (x))
        {
@@ -5622,7 +5632,6 @@ syms_of_comp (void)
   DEFSYM (Qfixnum, "fixnum");
   DEFSYM (Qscratch, "scratch");
   DEFSYM (Qlate, "late");
-  DEFSYM (Q__lambda_fixup, "--lambda-fixup");
   DEFSYM (Qgccjit, "gccjit");
   DEFSYM (Qcomp_subr_trampoline_install, "comp-subr-trampoline-install");
   DEFSYM (Qnative_comp_warning_on_missing_source,
@@ -5804,6 +5813,10 @@ syms_of_comp (void)
 verification of the native compiler.  */);
   comp_sanitizer_active = false;
 
+  DEFVAR_LISP ("comp--#$", Vcomp__hashdollar,
+    doc: /* Special value which will print as "#$".  */);
+  Vcomp__hashdollar = build_string ("#$");
+
   Fprovide (intern_c_string ("native-compile"), Qnil);
 #endif /* #ifdef HAVE_NATIVE_COMP */
 
diff --git a/src/pdumper.c b/src/pdumper.c
index d45bbc84bba..a27df8d96d4 100644
--- a/src/pdumper.c
+++ b/src/pdumper.c
@@ -5504,7 +5504,7 @@ dump_do_dump_relocation (const uintptr_t dump_base,
            XSETSUBR (tem, subr);
            Lisp_Object *fixup =
              &(comp_u->data_relocs[XFIXNUM (lambda_data_idx)]);
-           eassert (EQ (*fixup, Q__lambda_fixup));
+           eassert (EQ (*fixup, Vcomp__hashdollar));
            *fixup = tem;
            Fputhash (tem, Qt, comp_u->lambda_gc_guard_h);
          }