[ELF] Disallow input section description without a filename

[MaskRay]

Summary

GNU ld does not allow .foo : { (*foo) }, but we may recognize it as three
input section descriptions: file "(" with any section name, file "*foo" with
any section name, file ")" with any section name. Disallow the error-prone usage.

[peter.smith]

LGTM. GNU ld documentation for InputSection https://sourceware.org/binutils/docs/ld/Input-Section-Basics.html states that An input section description consists of a file name optionally followed by a list of section names in parentheses. I couldn't find anything about which characters in filenames were not permitted. It looks like both lld and gnu ld can handle these when quoted. For example for a file "file(o)" we can write { "file(o)"(.text) } I do hope parentheses in filenames are not common though. Maybe worth a test case if we've not got one already.

[MaskRay]

In D125523#3510971, @peter.smith wrote:

LGTM. GNU ld documentation for InputSection https://sourceware.org/binutils/docs/ld/Input-Section-Basics.html states that An input section description consists of a file name optionally followed by a list of section names in parentheses. I couldn't find anything about which characters in filenames were not permitted. It looks like both lld and gnu ld can handle these when quoted. For example for a file "file(o)" we can write { "file(o)"(.text) } I do hope parentheses in filenames are not common though. Maybe worth a test case if we've not got one already.

Most printable characters are accepted in the filename pattern, with exception to ()#%. Seems that disallowing ( and )` is sufficient for us to disambiguate the grammar.
Parentheses in filenames are accepted if quoted. Let me check whether there is an existing test...

lld/ELF/ScriptParser.cpp

if (tok == ";") {
       // https://sourceware.org/binutils/docs/ld/Output-Section-Data.html
       if (peek() != "(")
         setError("( expected, but got " + peek());
       osec->filler = readFill();
     } else if (tok == "SORT") {
       readSort();
     } else if (tok == "INCLUDE") {
       readInclude();
+    } else if (tok == "(" || tok == ")") {
+      setError("expected filename pattern");
     } else if (peek() == "(") {
       osec->commands.push_back(readInputSectionDescription(tok));
     } else {
       // We have a file name and no input sections description. It is not a
       // commonly used syntax, but still acceptable. In that case, all sections
       // from the file will be included.
       // FIXME: GNU ld permits INPUT_SECTION_FLAGS to be used here. We do not
       // handle this case here as it will already have been matched by the

lld/test/ELF/linkerscript/no-filename-spec.s

 # REQUIRES: x86
+# RUN: rm -rf %t && split-file %s %t
 # RUN: echo '.section .bar, "a"; .quad 1;' | \
 # RUN:   llvm-mc -filetype=obj -triple=x86_64-pc-linux - -o %tfile1.o
 # RUN: echo '.section .zed, "a"; .quad 2;' | \
 # RUN:   llvm-mc -filetype=obj -triple=x86_64-pc-linux - -o %tfile2.o
 
 ## We have a file name and no input sections description. In that case, all
 ## sections from the file specified should be included. Check that.
-# RUN: ld.lld -o %t --script %s %tfile1.o %tfile2.o
-# RUN: llvm-objdump -s %t | FileCheck %s
+# RUN: ld.lld -o %t/a -T %t/a.t %tfile1.o %tfile2.o
+# RUN: llvm-objdump -s %t/a | FileCheck %s
 
 # CHECK:      Contents of section .foo:
 # CHECK-NEXT:  01000000 00000000 02000000 00000000
 
+# RUN: not ld.lld -o /dev/null -T %t/b.t %tfile1.o 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not ld.lld -o /dev/null -T %t/c.t %tfile1.o 2>&1 | FileCheck %s --check-prefix=ERR
+# RUN: not ld.lld -o /dev/null -T %t/d.t %tfile1.o 2>&1 | FileCheck %s --check-prefix=ERR
+
+# ERR: error: {{.*}}.t:1: expected filename pattern
+
+#--- a.t
 SECTIONS {
  .foo : { *file1.o *file2.o }
 }
+
+#--- b.t
+SECTIONS { .foo : { (*foo) } }
+
+#--- c.t
+SECTIONS { .foo : { (*(foo)) } }
+
+#--- d.t
+SECTIONS { .foo : { )(*foo) } }