From: Carlos O'Donell <carlos@redhat.com>
To: "H.J. Lu" <hjl.tools@gmail.com>, libc-alpha@sourceware.org
Subject: Re: [PATCH v6] x86_64: Correct THREAD_SETMEM/THREAD_SETMEM_NC for movq [BZ #27591]
Date: Thu, 1 Apr 2021 09:51:44 -0400 [thread overview]
Message-ID: <2f275d3b-d3eb-cf6b-fce0-c2085b4c3eb3@redhat.com> (raw)
In-Reply-To: <20210319183536.1663081-1-hjl.tools@gmail.com>
On 3/19/21 2:35 PM, H.J. Lu via Libc-alpha wrote:
> config/i386/constraints.md in GCC has
>
> (define_constraint "e"
> "32-bit signed integer constant, or a symbolic reference known
> to fit that range (for immediate operands in sign-extending x86-64
> instructions)."
> (match_operand 0 "x86_64_immediate_operand"))
>
> Since movq takes a signed 32-bit immediate or a register source operand,
> use "er", instead of "nr"/"ir", constraint for 32-bit signed integer
> constant or register on movq.
v6, and this version looks amazing, much better than before and I think
we've arrived a good solution to this problem. Thank you for your efforts
and working through the details.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
> ---
> sysdeps/x86_64/Makefile | 2 +
> sysdeps/x86_64/nptl/tls.h | 10 ++++-
> sysdeps/x86_64/tst-x86-64-tls-1.c | 64 +++++++++++++++++++++++++++++++
> 3 files changed, 74 insertions(+), 2 deletions(-)
> create mode 100644 sysdeps/x86_64/tst-x86-64-tls-1.c
>
> diff --git a/sysdeps/x86_64/Makefile b/sysdeps/x86_64/Makefile
> index d1d7cb9d2e..06a444b6af 100644
> --- a/sysdeps/x86_64/Makefile
> +++ b/sysdeps/x86_64/Makefile
> @@ -183,6 +183,8 @@ ifeq (no,$(build-hardcoded-path-in-tests))
> tests-container += tst-glibc-hwcaps-cache
> endif
>
> +tests-internal += tst-x86-64-tls-1
OK. Test is internal with use of tls.h header.
> +
> endif # $(subdir) == elf
>
> ifeq ($(subdir),csu)
> diff --git a/sysdeps/x86_64/nptl/tls.h b/sysdeps/x86_64/nptl/tls.h
> index 20f0958780..a78c4f4d01 100644
> --- a/sysdeps/x86_64/nptl/tls.h
> +++ b/sysdeps/x86_64/nptl/tls.h
> @@ -271,8 +271,11 @@ _Static_assert (offsetof (tcbhead_t, __glibc_unused2) == 0x80,
> "i" (offsetof (struct pthread, member))); \
> else /* 8 */ \
> { \
> + /* Since movq takes a signed 32-bit immediate or a register source \
> + operand, use "er" constraint for 32-bit signed integer constant \
> + or register. */ \
> asm volatile ("movq %q0,%%fs:%P1" : \
> - : IMM_MODE ((uint64_t) cast_to_integer (value)), \
> + : "er" ((uint64_t) cast_to_integer (value)), \
OK. Great change, very clear, and uses immediate or register to get it to work.
> "i" (offsetof (struct pthread, member))); \
> }})
>
> @@ -296,8 +299,11 @@ _Static_assert (offsetof (tcbhead_t, __glibc_unused2) == 0x80,
> "r" (idx)); \
> else /* 8 */ \
> { \
> + /* Since movq takes a signed 32-bit immediate or a register source \
> + operand, use "er" constraint for 32-bit signed integer constant \
> + or register. */ \
> asm volatile ("movq %q0,%%fs:%P1(,%q2,8)" : \
> - : IMM_MODE ((uint64_t) cast_to_integer (value)), \
> + : "er" ((uint64_t) cast_to_integer (value)), \
> "i" (offsetof (struct pthread, member[0])), \
OK.
> "r" (idx)); \
> }})
> diff --git a/sysdeps/x86_64/tst-x86-64-tls-1.c b/sysdeps/x86_64/tst-x86-64-tls-1.c
> new file mode 100644
> index 0000000000..354635884e
> --- /dev/null
> +++ b/sysdeps/x86_64/tst-x86-64-tls-1.c
> @@ -0,0 +1,64 @@
> +/* Test THREAD_SETMEM and THREAD_SETMEM_NC for IMM64.
> + Copyright (C) 2021 Free Software Foundation, Inc.
> + This file is part of the GNU C Library.
> +
> + The GNU C Library is free software; you can redistribute it and/or
> + modify it under the terms of the GNU Lesser General Public
> + License as published by the Free Software Foundation; either
> + version 2.1 of the License, or (at your option) any later version.
> +
> + The GNU C Library is distributed in the hope that it will be useful,
> + but WITHOUT ANY WARRANTY; without even the implied warranty of
> + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
> + Lesser General Public License for more details.
> +
> + You should have received a copy of the GNU Lesser General Public
> + License along with the GNU C Library; if not, see
> + <https://www.gnu.org/licenses/>. */
> +
> +#include <tls.h>
> +#include <support/check.h>
> +
> +static int
> +do_test (void)
> +{
> + unsigned long long int saved_ssp_base, ssp_base;
> + saved_ssp_base = THREAD_GETMEM (THREAD_SELF, header.ssp_base);
> +
> + THREAD_SETMEM (THREAD_SELF, header.ssp_base, (1ULL << 57) - 1);
> + ssp_base = THREAD_GETMEM (THREAD_SELF, header.ssp_base);
> + if (ssp_base != ((1ULL << 57) - 1))
> + FAIL_EXIT1 ("THREAD_SETMEM: 0x%llx != 0x%llx",
> + ssp_base, (1ULL << 57) - 1);
> +
> + THREAD_SETMEM (THREAD_SELF, header.ssp_base, -1ULL);
> + ssp_base = THREAD_GETMEM (THREAD_SELF, header.ssp_base);
> + if (ssp_base != -1ULL)
> + FAIL_EXIT1 ("THREAD_SETMEM: 0x%llx != 0x%llx", ssp_base, -1ULL);
> +
> + THREAD_SETMEM (THREAD_SELF, header.ssp_base, saved_ssp_base);
> +#ifndef __ILP32__
> + struct pthread_key_data *saved_specific, *specific;
> + saved_specific = THREAD_GETMEM_NC (THREAD_SELF, specific, 1);
> +
> + uintptr_t value = (1UL << 57) - 1;
> + THREAD_SETMEM_NC (THREAD_SELF, specific, 1,
> + (struct pthread_key_data *) value);
> + specific = THREAD_GETMEM_NC (THREAD_SELF, specific, 1);
> + if (specific != (struct pthread_key_data *) value)
> + FAIL_EXIT1 ("THREAD_GETMEM_NC: %p != %p",
> + specific, (struct pthread_key_data *) value);
> +
> + THREAD_SETMEM_NC (THREAD_SELF, specific, 1,
> + (struct pthread_key_data *) -1UL);
> + specific = THREAD_GETMEM_NC (THREAD_SELF, specific, 1);
> + if (specific != (struct pthread_key_data *) -1UL)
> + FAIL_EXIT1 ("THREAD_GETMEM_NC: %p != %p",
> + specific, (struct pthread_key_data *) -1UL);
> +
> + THREAD_SETMEM_NC (THREAD_SELF, specific, 1, saved_specific);
> +#endif
> + return 0;
> +}
> +
> +#include <support/test-driver.c>
OK.
>
--
Cheers,
Carlos.
next prev parent reply other threads:[~2021-04-01 13:51 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-03-19 18:35 H.J. Lu
2021-04-01 13:51 ` Carlos O'Donell [this message]
-- strict thread matches above, loose matches on Subject: below --
2021-02-02 19:12 [PATCH] x86_64: Update THREAD_SETMEM/THREAD_SETMEM_NC for IMM64 H.J. Lu
2021-03-01 13:30 ` Carlos O'Donell
2021-03-02 14:21 ` [PATCH v2] " H.J. Lu
2021-03-08 22:28 ` Carlos O'Donell
2021-03-09 0:09 ` [PATCH v3] " H.J. Lu
2021-03-15 12:49 ` Carlos O'Donell
2021-03-15 13:29 ` [PATCH v4] " H.J. Lu
2021-03-16 3:01 ` Carlos O'Donell
2021-03-16 9:04 ` Andreas Schwab
2021-03-16 15:45 ` [PATCH v5] x86_64: Update THREAD_SETMEM/THREAD_SETMEM_NC for IMM64 [BZ #27591] H.J. Lu
2021-03-16 16:02 ` Andreas Schwab
2021-03-16 16:10 ` Florian Weimer
2021-03-16 16:34 ` Carlos O'Donell
2021-03-16 16:50 ` [PATCH v6] x86_64: Correct THREAD_SETMEM/THREAD_SETMEM_NC for movq " H.J. Lu
2021-03-16 16:52 ` H.J. Lu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2f275d3b-d3eb-cf6b-fce0-c2085b4c3eb3@redhat.com \
--to=carlos@redhat.com \
--cc=hjl.tools@gmail.com \
--cc=libc-alpha@sourceware.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).