This is the mail archive of the binutils@sources.redhat.com mailing list for the binutils 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: demand_empty_rest_of_line and ignore_rest_of_line


Richard Earnshaw <Richard.Earnshaw@arm.com> writes:

> > Currently (gcc 3.4.0) there are 5 targets that actually make use of
> > #NO_APP: arm, cris, m68k, ns32k and vax (and m68k-linux didn't in gcc
> > 3.0.x upto 3.3.x due to a side effect of including "elfos.h", which
> > redefines ASM_FILE_START).
> 
> That's probably misleading.  The gcc might emit a directive for that,
> but I'm almost certain the ARM assembler for one takes no notice of it.

The code to handle #NO_APP in gas is independent of the CPU backend.
The initial #NO_APP comment is recognized in input_file_open(); it
must be the first characters in the file.  It then turns off
preprocessing (i.e., calls to do_scrub_chars) except for code which
follows #APP up to #NO_APP.

That said, I don't see #NO_APP in assembler code generated by an
xscale-elf compiler.  In fact, I see that
TARGET_ASM_FILE_START_APP_OFF is defined to be true, which should
generate #NO_APP, but that ASM_APP_OFF is defined to be "", so the
effect is nullified.  That is, the ARM compiler does not generally use
#NO_APP.

Ian


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