On 17/07/2026 15:32, Alice Carlotti wrote:
> On Fri, Jul 17, 2026 at 12:26:08PM +0000, Richard Earnshaw via Sourceware
> Forge wrote:
> > From: Richard Earnshaw <[email protected]>
> >
> > Arm has sunset the TME extension from the aarch64 architecture since
> > there are no known implementations. Since this has been present in the
> > compiler for a while though, we need to go through a deprecation cycle.
> >
> > This patch starts that process by:
> > - removing it from the documentation
> > - warning if the option is used during compilation
> > - removing the tests for it.
> >
> > gcc/ChangeLog:
> >
> > * config/aarch64/aarch64.cc (aarch64_override_options): Warn if
> > +tme is used as an architecture feature
> > (aarch64_process_target_attr): Similarly, but only if TME
> > was previously disabled.
> > * doc/invoke.texi (aarch64): Remove mentions of +tme.
> >
> > gcc/testsuite/ChangeLog:
> >
> > * gcc.target/aarch64/pragma_cpp_predefs_2.c: Drop tests for TME.
> > * gcc.target/aarch64/acle/tme.c: Removed.
> > * gcc.target/aarch64/acle/tme_guard-1.c: Removed.
> > * gcc.target/aarch64/acle/tme_guard-2.c: Removed.
> > * gcc.target/aarch64/acle/tme_guard-3.c: Removed.
> > * gcc.target/aarch64/acle/tme_guard-4.c: Removed.
> > ---
> > gcc/config/aarch64/aarch64.cc | 10 +++++-
> > gcc/doc/invoke.texi | 2 --
> > gcc/testsuite/gcc.target/aarch64/acle/tme.c | 34 -------------------
> > .../gcc.target/aarch64/acle/tme_guard-1.c | 9 -----
> > .../gcc.target/aarch64/acle/tme_guard-2.c | 10 ------
> > .../gcc.target/aarch64/acle/tme_guard-3.c | 9 -----
> > .../gcc.target/aarch64/acle/tme_guard-4.c | 10 ------
> > .../gcc.target/aarch64/pragma_cpp_predefs_2.c | 12 -------
> > 8 files changed, 9 insertions(+), 87 deletions(-)
> > delete mode 100644 gcc/testsuite/gcc.target/aarch64/acle/tme.c
> > delete mode 100644 gcc/testsuite/gcc.target/aarch64/acle/tme_guard-1.c
> > delete mode 100644 gcc/testsuite/gcc.target/aarch64/acle/tme_guard-2.c
> > delete mode 100644 gcc/testsuite/gcc.target/aarch64/acle/tme_guard-3.c
> > delete mode 100644 gcc/testsuite/gcc.target/aarch64/acle/tme_guard-4.c
> >
> > diff --git a/gcc/config/aarch64/aarch64.cc b/gcc/config/aarch64/aarch64.cc
> > index a1f91dd425ef..2d34ee09b5be 100644
> > --- a/gcc/config/aarch64/aarch64.cc
> > +++ b/gcc/config/aarch64/aarch64.cc
> > @@ -20428,6 +20428,10 @@ aarch64_override_options (void)
> > if (TARGET_ILP32)
> > warning (OPT_Wdeprecated, "%<-mabi=ilp32%> is deprecated");
> >
> > + if (TARGET_TME)
> > + warning (OPT_Wdeprecated,
> > + "the architecture extension %<+tme%> is deprecated");
> > +
> > if (global_options.x_pcrelative_literal_loads == 1)
> > warning (OPT_Wdeprecated, "%<-mpc-relative-literal-loads%> is
> > deprecated");
> >
> > @@ -21078,6 +21082,11 @@ aarch64_process_target_attr (tree args)
> > while (token)
> > {
> > num_attrs++;
> > +
> > + if ((strcmp (token, "+tme") == 0) && !TARGET_TME)
> > + warning (OPT_Wdeprecated,
> > + "the architecture extension %<+tme%> is deprecated");
> > +
>
> This check misses stuff like "+memtag+tme", and I think it's also in the wrong
> place. The ideal place to check would probably be in aarch64_parse_extension;
> if its use in the driver makes that difficult, then we should find some way of
> passing the information out of that function to somewhere that can emit the
> warning.
Good catch, I forgot you can (of course) have multiple extensions
separated by + here. Thanks for spotting that.
>
> We could also use a global variable to ensure we only warn once about each
> deprecated feature - if there's multiple instances, users can easily search
> the
> code for "+tme" to find all of them.
While I agree that we definitely don't want multiple diagnostics at the
same location, I believe it is conventional for diagnostics to warn
about all instances of a problem, so that users can fix them in bulk
without having to go through an edit-compile cycle for every fix.
Alex
>
> Alice
>
> > if (!aarch64_process_one_target_attr (token))
> > {
> > /* Check if token is possibly an arch extension without
> > @@ -21094,7 +21103,6 @@ aarch64_process_target_attr (tree args)
> > error ("pragma or attribute %<target(\"%s\")%> is not valid",
> > token);
> > return false;
> > }
> > -
> > token = strtok_r (NULL, ",", &str_to_check);
> > }
> >
