From: Samuel Thibault <samuel.thibault@gnu.org>
To: Sergey Bugaev <bugaevc@gmail.com>
Cc: libc-alpha@sourceware.org, bug-hurd@gnu.org
Subject: Re: [RFC PATCH glibc v2 26/34] hurd: Remove __hurd_local_reply_port
Date: Fri, 14 Apr 2023 19:33:32 +0200 [thread overview]
Message-ID: <20230414173332.afm47w6clabzklex@begin> (raw)
In-Reply-To: <20230413115812.267158-1-bugaevc@gmail.com>
Applied with fixing the __mig_dealloc_reply_port(NULL) cases, thanks!
Sergey Bugaev, le jeu. 13 avril 2023 14:58:12 +0300, a ecrit:
> Now that the signal code no longer accesses it, the only real user of it
> was mig-reply.c, so move the logic for managing the port there.
>
> If we're in SHARED and outside of rtld, we know that __LIBC_NO_TLS ()
> always evaluates to 0, and a TLS reply port will always be used, not
> __hurd_reply_port0. Still, the compiler does not see that
> __hurd_reply_port0 is never used due to its address being taken. To deal
> with this, explicitly compile out __hurd_reply_port0 when we know we
> won't use it.
>
> Also, instead of accessing the port via THREAD_SELF->reply_port, this
> uses THREAD_GETMEM and THREAD_SETMEM directly, avoiding possible
> miscompilations.
>
> Signed-off-by: Sergey Bugaev <bugaevc@gmail.com>
> ---
> Changes since v1:
> * Only deallocate the port in __mig_dealloc_reply_port if it's the same
> as the stored port, because of signals.
> * Add a comment detailing the reasoning.
> * Add two assertions: that ARG and PORT are exactly the same (unless PORT
> has been invalidated), and that we succeed deallocating PORT.
>
> NOTE: Completely untested! Do your own testing! You have been warned!
>
> hurd/hurd/threadvar.h | 7 ----
> sysdeps/mach/hurd/mig-reply.c | 70 +++++++++++++++++++++++++++++------
> 2 files changed, 58 insertions(+), 19 deletions(-)
>
> diff --git a/hurd/hurd/threadvar.h b/hurd/hurd/threadvar.h
> index f5c6a278..c476d988 100644
> --- a/hurd/hurd/threadvar.h
> +++ b/hurd/hurd/threadvar.h
> @@ -29,11 +29,4 @@
> extern unsigned long int __hurd_sigthread_stack_base;
> extern unsigned long int __hurd_sigthread_stack_end;
>
> -/* Store the MiG reply port reply port until we enable TLS. */
> -extern mach_port_t __hurd_reply_port0;
> -
> -/* This returns either the TLS reply port variable, or a single-thread variable
> - when TLS is not initialized yet. */
> -#define __hurd_local_reply_port (*(__LIBC_NO_TLS () ? &__hurd_reply_port0 : &THREAD_SELF->reply_port))
> -
> #endif /* hurd/threadvar.h */
> diff --git a/sysdeps/mach/hurd/mig-reply.c b/sysdeps/mach/hurd/mig-reply.c
> index 33ff260e..3fdee80e 100644
> --- a/sysdeps/mach/hurd/mig-reply.c
> +++ b/sysdeps/mach/hurd/mig-reply.c
> @@ -17,22 +17,46 @@
>
> #include <mach.h>
> #include <mach/mig_support.h>
> -#include <hurd/threadvar.h>
> +#include <tls.h>
>
> /* These functions are called by MiG-generated code. */
>
> +#if !defined (SHARED) || IS_IN (rtld)
> mach_port_t __hurd_reply_port0;
> +#endif
> +
> +static mach_port_t
> +get_reply_port (void)
> +{
> +#if !defined (SHARED) || IS_IN (rtld)
> + if (__LIBC_NO_TLS ())
> + return __hurd_reply_port0;
> +#endif
> + return THREAD_GETMEM (THREAD_SELF, reply_port);
> +}
> +
> +static void
> +set_reply_port (mach_port_t port)
> +{
> +#if !defined (SHARED) || IS_IN (rtld)
> + if (__LIBC_NO_TLS ())
> + __hurd_reply_port0 = port;
> + else
> +#endif
> + THREAD_SETMEM (THREAD_SELF, reply_port, port);
> +}
>
> /* Called by MiG to get a reply port. */
> mach_port_t
> __mig_get_reply_port (void)
> {
> - if (__hurd_local_reply_port == MACH_PORT_NULL
> - || (&__hurd_local_reply_port != &__hurd_reply_port0
> - && __hurd_local_reply_port == __hurd_reply_port0))
> - __hurd_local_reply_port = __mach_reply_port ();
> -
> - return __hurd_local_reply_port;
> + mach_port_t port = get_reply_port ();
> + if (__glibc_unlikely (port == MACH_PORT_NULL))
> + {
> + port = __mach_reply_port ();
> + set_reply_port (port);
> + }
> + return port;
> }
> weak_alias (__mig_get_reply_port, mig_get_reply_port)
> libc_hidden_def (__mig_get_reply_port)
> @@ -41,12 +65,34 @@ libc_hidden_def (__mig_get_reply_port)
> void
> __mig_dealloc_reply_port (mach_port_t arg)
> {
> - mach_port_t port = __hurd_local_reply_port;
> - __hurd_local_reply_port = MACH_PORT_NULL; /* So the mod_refs RPC won't use it. */
> + error_t err;
> + mach_port_t port = get_reply_port ();
> +
> + set_reply_port (MACH_PORT_NULL); /* So the mod_refs RPC won't use it. */
> +
> + /* Normally, ARG should be the same as PORT that we store. However, if a
> + signal has interrupted the RPC, the stored PORT has been deallocated and
> + reset to MACH_PORT_NULL (or possibly MACH_PORT_DEAD). In this case the
> + MIG routine still has the old name, which it passes to us here. We must
> + not deallocate (or otherwise touch) it, since it may be already allocated
> + to another port right. Fortunately MIG itself doesn't do anything with
> + the reply port on errors either, other than immediately calling this
> + function.
> +
> + And so:
> + 1. Assert that things are sane, i.e. and PORT is either invalid or same
> + as ARG.
> + 2. Only deallocate the name if our stored PORT still names it. In that
> + case we're sure the right has not been deallocated / the name reused.
> + */
> +
> + if (!MACH_PORT_VALID (port))
> + return;
> + assert (port == arg);
>
> - if (MACH_PORT_VALID (port))
> - __mach_port_mod_refs (__mach_task_self (), port,
> - MACH_PORT_RIGHT_RECEIVE, -1);
> + err = __mach_port_mod_refs (__mach_task_self (), port,
> + MACH_PORT_RIGHT_RECEIVE, -1);
> + assert_perror (err);
> }
> weak_alias (__mig_dealloc_reply_port, mig_dealloc_reply_port)
> libc_hidden_def (__mig_dealloc_reply_port)
> --
> 2.39.2
>
--
Samuel
---
Pour une évaluation indépendante, transparente et rigoureuse !
Je soutiens la Commission d'Évaluation de l'Inria.
next prev parent reply other threads:[~2023-04-14 17:33 UTC|newest]
Thread overview: 140+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-03-19 15:09 [RFC PATCH 00/34] The rest of the x86_64-gnu port Sergey Bugaev
2023-03-19 15:09 ` [RFC PATCH gnumach 01/34] Add i386_fsgs_base_state Sergey Bugaev
2023-04-02 22:43 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH gnumach 02/34] Remove bootstrap.defs Sergey Bugaev
2023-04-02 22:43 ` Samuel Thibault
2023-04-03 9:39 ` Sergey Bugaev
2023-03-19 15:09 ` [RFC PATCH gnumach 03/34] Make exception subcode a long Sergey Bugaev
2023-04-02 22:45 ` Samuel Thibault
2023-04-03 9:32 ` Sergey Bugaev
2023-04-06 2:11 ` Flávio Cruz
2023-04-10 23:52 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 04/34] hurd: " Sergey Bugaev
2023-04-02 22:52 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 05/34] hurd: Remove __hurd_threadvar_stack_{offset,mask} Sergey Bugaev
2023-04-02 22:53 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 06/34] hurd: Swap around two function calls Sergey Bugaev
2023-04-02 22:54 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 07/34] hurd: Fix file name in #error Sergey Bugaev
2023-04-02 22:55 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 08/34] hurd: Disable O_TRUNC and FS_RETRY_MAGICAL in rtld Sergey Bugaev
2023-04-02 22:57 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 09/34] hurd: Fix _hurd_setup_sighandler () signature Sergey Bugaev
2023-04-02 22:58 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 10/34] stdio-common: Fix building when !IS_IN (libc) Sergey Bugaev
2023-04-02 23:01 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 11/34] mach, hurd: Drop __libc_lock_self0 Sergey Bugaev
2023-04-02 23:02 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 12/34] hurd: More 64-bit integer casting fixes Sergey Bugaev
2023-04-02 23:03 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 13/34] x86-64: Disable prefer_map_32bit_exec tunable on non-Linux Sergey Bugaev
2023-04-02 23:09 ` Samuel Thibault
2023-04-03 10:10 ` Sergey Bugaev
2023-04-03 19:02 ` H.J. Lu
2023-04-03 20:11 ` Sergey Bugaev
2023-03-19 15:09 ` [RFC PATCH glibc 14/34] hurd: Move rtld-strncpy-c.c out of mach/hurd/ Sergey Bugaev
2023-04-02 23:10 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 15/34] hurd: Use uintptr_t for register values in trampoline.c Sergey Bugaev
2023-04-02 23:13 ` Samuel Thibault
2023-03-19 15:09 ` [RFC PATCH glibc 16/34] hurd: Add sys/ucontext.h and sigcontext.h for x86_64 Sergey Bugaev
2023-04-10 18:39 ` Samuel Thibault
2023-04-10 19:07 ` Sergey Bugaev
2023-04-10 19:21 ` Samuel Thibault
2023-04-10 18:58 ` Samuel Thibault
2023-04-10 19:13 ` Sergey Bugaev
2023-04-10 19:21 ` Samuel Thibault
2023-04-10 21:50 ` Sergey Bugaev
2023-04-10 22:23 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 17/34] hurd: Implement x86_64/intr-msg.h Sergey Bugaev
2023-04-10 18:41 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 18/34] hurd: Port trampoline.c to x86_64 Sergey Bugaev
2023-04-03 11:56 ` [PATCH v2 18.0/34] Alignment-respecting x86_64 trampoline.c Sergey Bugaev
2023-04-03 11:56 ` [PATCH v2 18.1/34] hurd: Do not declare local variables volatile Sergey Bugaev
2023-04-10 18:42 ` Samuel Thibault
2023-04-03 11:56 ` [PATCH v2 18.2/34] hurd: Port trampoline.c to x86_64 Sergey Bugaev
2023-04-10 19:04 ` Samuel Thibault
2023-04-10 21:33 ` Sergey Bugaev
2023-03-19 15:10 ` [RFC PATCH glibc 19/34] hurd: Move a couple of singal-related files to x86 Sergey Bugaev
2023-04-02 23:15 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 20/34] htl: Add tcb-offsets.sym for x86_64 Sergey Bugaev
2023-04-02 23:16 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 21/34] x86_64: Add rtld-stpncpy & rtld-strncpy Sergey Bugaev
2023-04-02 23:18 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 22/34] htl: Implement thread_set_pcsptp for x86_64 Sergey Bugaev
2023-04-02 23:19 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 23/34] elf: Stop including tls.h in ldsodefs.h Sergey Bugaev
2023-04-02 23:20 ` Samuel Thibault
2023-04-03 9:26 ` Sergey Bugaev
2023-04-10 21:26 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 24/34] hurd: Only check for TLS initialization inside rtld or in static builds Sergey Bugaev
2023-04-10 21:33 ` Samuel Thibault
2023-04-11 18:57 ` Samuel Thibault
2023-04-11 19:18 ` Samuel Thibault
2023-04-11 20:03 ` Samuel Thibault
2023-04-11 20:27 ` Sergey Bugaev
2023-04-11 21:23 ` Samuel Thibault
2023-04-12 8:36 ` Sergey Bugaev
2023-04-12 9:00 ` Samuel Thibault
2023-04-12 10:42 ` Sergey Bugaev
2023-04-12 10:45 ` Samuel Thibault
2023-04-12 17:18 ` Sergey Bugaev
2023-04-12 23:46 ` Samuel Thibault
2023-04-13 10:02 ` Sergey Bugaev
2023-04-13 10:10 ` Samuel Thibault
2023-04-13 12:17 ` Sergey Bugaev
2023-04-13 21:47 ` Samuel Thibault
2023-04-13 22:21 ` Samuel Thibault
2023-04-14 8:29 ` Sergey Bugaev
2023-04-14 8:36 ` Samuel Thibault
2023-04-14 8:53 ` Sergey Bugaev
2023-04-14 9:09 ` Samuel Thibault
2023-04-14 9:23 ` Sergey Bugaev
2023-04-14 9:31 ` Samuel Thibault
2023-04-17 7:16 ` Samuel Thibault
2023-04-14 17:34 ` Samuel Thibault
2023-04-14 19:52 ` Sergey Bugaev
2023-03-19 15:10 ` [RFC PATCH glibc 25/34] hurd: Improve reply port handling when exiting signal handlers Sergey Bugaev
2023-04-10 22:03 ` Samuel Thibault
2023-04-11 7:44 ` Sergey Bugaev
2023-04-11 20:15 ` Samuel Thibault
2023-04-11 20:35 ` Sergey Bugaev
2023-04-12 22:54 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 26/34] hurd: Remove __hurd_local_reply_port Sergey Bugaev
2023-04-10 22:07 ` Samuel Thibault
2023-04-10 22:35 ` Samuel Thibault
2023-04-11 8:00 ` Sergey Bugaev
2023-04-11 20:18 ` Samuel Thibault
2023-04-13 11:58 ` [RFC PATCH glibc v2 " Sergey Bugaev
2023-04-13 13:12 ` Samuel Thibault
2023-04-13 13:20 ` Sergey Bugaev
2023-04-13 21:28 ` Samuel Thibault
2023-04-14 17:33 ` Samuel Thibault [this message]
2023-04-14 20:29 ` Sergey Bugaev
2023-04-15 6:45 ` Samuel Thibault
2023-04-15 7:34 ` Sergey Bugaev
2023-04-15 7:42 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 27/34] hurd: Don't leak __hurd_reply_port0 Sergey Bugaev
2023-04-10 22:25 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 28/34] hurd: Implement _hurd_longjmp_thread_state for x86_64 Sergey Bugaev
2023-04-02 23:23 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 29/34] hurd: Add vm_param.h " Sergey Bugaev
2023-04-02 23:24 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 30/34] hurd: Implement longjmp " Sergey Bugaev
2023-03-19 15:10 ` [RFC PATCH glibc 31/34] hurd: Microoptimize _hurd_self_sigstate () Sergey Bugaev
2023-04-02 23:26 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 32/34] hurd: Implement sigreturn for x86_64 Sergey Bugaev
2023-04-03 11:47 ` [PATCH v2] " Sergey Bugaev
2023-03-19 15:10 ` [RFC PATCH glibc 33/34] hurd: Create abilist files for lib{mach,hurd}user Sergey Bugaev
2023-03-19 15:19 ` Samuel Thibault
2023-03-19 15:39 ` Sergey Bugaev
2023-03-19 15:43 ` Samuel Thibault
2023-03-19 15:10 ` [RFC PATCH glibc 34/34] hurd: Add expected abilist files for x86_64 Sergey Bugaev
2023-03-19 18:04 ` Florian Weimer
2023-03-19 20:14 ` [PATCH v2] " Sergey Bugaev
2023-03-20 6:30 ` Florian Weimer
2023-03-19 16:44 ` [RFC PATCH 00/34] The rest of the x86_64-gnu port Luca
2023-03-20 5:03 ` Flávio Cruz
2023-04-02 23:30 ` Samuel Thibault
2023-04-10 19:20 ` Samuel Thibault
2023-04-10 21:24 ` Sergey Bugaev
2023-04-10 21:27 ` Samuel Thibault
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=20230414173332.afm47w6clabzklex@begin \
--to=samuel.thibault@gnu.org \
--cc=bug-hurd@gnu.org \
--cc=bugaevc@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).