From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from eggs.gnu.org (eggs.gnu.org [IPv6:2001:470:142:3::10]) by sourceware.org (Postfix) with ESMTPS id BBDFF3858C5E for ; Mon, 10 Apr 2023 18:41:43 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org BBDFF3858C5E Authentication-Results: sourceware.org; dmarc=pass (p=none dis=none) header.from=gnu.org Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=gnu.org Received: from fencepost.gnu.org ([2001:470:142:3::e]) by eggs.gnu.org with esmtps (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1plwST-0006Yg-NQ; Mon, 10 Apr 2023 14:41:41 -0400 DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=gnu.org; s=fencepost-gnu-org; h=In-Reply-To:MIME-Version:References:Subject:To:From: Date; bh=0nIPGMAIRCksJBrAooiQE3NdmLU+qq/NxRbvp+1PJYY=; b=dhHlyw164SH7sFnKhU4Z Cc8CgoqrmozOH8Hf+vHrPWUnoBRF4qN/h6x29bSRjRRlbAWVcF1xLdPZnbvee9AjBkj/Dz7Hkt7aM qYJzDCaLA3YyfnVOnWkwE2t/KRrVczlWXCMtoQs/0OpfdGQ3y+4625heJKNWVIVZIGdr4BcQAlLNK IULSmW6FqYunweP5VH+JPKYx1wTfccQ3XUhZqfePObuqhgdsXJA9+1FmkTiLZ6qrvXDL/9XT+ggkf Crg8SBiYCsKwVW/pQn+K3+/Wqqrdk0VaiYYmpj9ryPpdmcdlUUlcodjgRFUznBcTKCV+5cncSZPdY AlyTEEFcYYqHFg==; Received: from [2a01:cb19:4a:a400:de41:a9ff:fe47:ec49] (helo=begin) by fencepost.gnu.org with esmtpsa (TLS1.2:ECDHE_RSA_AES_256_GCM_SHA384:256) (Exim 4.90_1) (envelope-from ) id 1plwSS-00059M-Oh; Mon, 10 Apr 2023 14:41:40 -0400 Received: from samy by begin with local (Exim 4.96) (envelope-from ) id 1plwSR-0000bP-2K; Mon, 10 Apr 2023 20:41:39 +0200 Date: Mon, 10 Apr 2023 20:41:39 +0200 From: Samuel Thibault To: Sergey Bugaev Cc: libc-alpha@sourceware.org, bug-hurd@gnu.org Subject: Re: [RFC PATCH glibc 17/34] hurd: Implement x86_64/intr-msg.h Message-ID: <20230410184139.cdkw6olzy2ruwnlf@begin> Mail-Followup-To: Sergey Bugaev , libc-alpha@sourceware.org, bug-hurd@gnu.org References: <20230319151017.531737-1-bugaevc@gmail.com> <20230319151017.531737-18-bugaevc@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <20230319151017.531737-18-bugaevc@gmail.com> Organization: I am not organized User-Agent: NeoMutt/20170609 (1.8.3) X-Spam-Status: No, score=-12.7 required=5.0 tests=BAYES_00,DKIMWL_WL_HIGH,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,GIT_PATCH_0,KAM_SHORT,SPF_HELO_PASS,SPF_PASS,TXREP autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: Applied, thanks! Sergey Bugaev, le dim. 19 mars 2023 18:10:00 +0300, a ecrit: > This is a prelimenary version of intr-msg.h. I can't know whether or not > it works until we can test it. Sure :) > The code relies on syscall preserving values > of the registers used to pass syscall arguments. The code in SYSCALL_EXAMINE > that just compares two bytes to the encoded syscall instruction is sure to > give many false positives; we're going to need to come up with a more reliable > scheme. Not necessarily: this is only used for msg_report_wait, which is already extra cautious about what it's doing, at worse it will report dumb information, which should be fine enough for msg_report_wait. Possibly we could filter out by addresses of the functions known to perform an actual system call. Samuel > sysdeps/mach/hurd/x86_64/intr-msg.h | 119 ++++++++++++++++++++++++++++ > 1 file changed, 119 insertions(+) > create mode 100644 sysdeps/mach/hurd/x86_64/intr-msg.h > > diff --git a/sysdeps/mach/hurd/x86_64/intr-msg.h b/sysdeps/mach/hurd/x86_64/intr-msg.h > new file mode 100644 > index 00000000..5d94c85f > --- /dev/null > +++ b/sysdeps/mach/hurd/x86_64/intr-msg.h > @@ -0,0 +1,119 @@ > +/* Machine-dependent details of interruptible RPC messaging. x86_64 version. > + Copyright (C) 1995-2023 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 > + . */ > + > + > +/* Note that we must mark OPTION and TIMEOUT as outputs of this operation, > + to indicate that the signal thread might mutate them as part > + of sending us to a signal handler. */ > + > +#define INTR_MSG_TRAP(msg, option, send_size, rcv_size, rcv_name, timeout, notify, cancel_p, intr_port_p) \ > +({ \ > + error_t err; \ > + register uintptr_t rcv_size_r10 asm ("r10") = rcv_size; \ > + register uintptr_t rcv_name_r8 asm ("r8") = rcv_name; \ > + register uintptr_t timeout_r9 asm ("r9") = timeout; \ > + asm volatile ( \ > + ".globl _hurd_intr_rpc_msg_about_to\n" \ > + ".globl _hurd_intr_rpc_msg_setup_done\n" \ > + ".globl _hurd_intr_rpc_msg_in_trap\n" \ > + /* Clear rax before we do the check for cancel below. This is to > + detect rax being set to non-zero (actually MACH_SEND_INTERRUPTED) > + from the outside (namely, _hurdsig_abort_rpcs), which signals us > + to skip the trap we were about to enter. */ \ > + " xor %[err], %[err]\n" \ > + "_hurd_intr_rpc_msg_about_to:" \ > + /* We need to make a last check of cancel, in case we got interrupted > + right before _hurd_intr_rpc_msg_about_to. */ \ > + " cmpl $0, %[cancel]\n" \ > + " jz _hurd_intr_rpc_msg_do\n" \ > + /* We got interrupted, note so and return EINTR. */ \ > + " movl $0, %[intr_port]\n" \ > + " movl %[eintr], %[err]\n" \ > + " jmp _hurd_intr_rpc_msg_sp_restored\n" \ > + "_hurd_intr_rpc_msg_do:" \ > + /* Ok, prepare the mach_msg_trap arguments. We pass the first 6 args > + in registers, and push the last one and a fake return address. */ \ > + " pushq %[notify]\n" \ > + " pushq $0\n" \ > + "_hurd_intr_rpc_msg_setup_done:" \ > + /* From here on, it is safe to make us jump over the syscall. Now > + check if we have been told to skip the syscall while running > + the above. */ \ > + " test %[err], %[err]\n" \ > + " jnz _hurd_intr_rpc_msg_in_trap\n" \ > + /* Do the actual syscall. */ \ > + " movq $-25, %%rax\n" \ > + "_hurd_intr_rpc_msg_do_trap: syscall # status in %[err]\n" \ > + "_hurd_intr_rpc_msg_in_trap:" \ > + /* Clean the arguments. */ \ > + " addq $16, %%rsp\n" \ > + "_hurd_intr_rpc_msg_sp_restored:" \ > + : [err] "=&a" (err), "+S" (option), [intr_port] "=m" (*intr_port_p), \ > + "+r" (timeout_r9) \ > + : "D" (msg), "d" (send_size), "r" (rcv_size_r10), "r" (rcv_name_r8), \ > + [notify] "rm" ((uintptr_t) notify), [cancel] "m" (*cancel_p), \ > + [eintr] "i" (EINTR) \ > + : "rcx", "r11"); \ > + timeout = timeout_r9; \ > + err; \ > +}) > + > +#include "hurdfault.h" > + > +/* This cannot be an inline function because it calls setjmp. */ > +#define SYSCALL_EXAMINE(state, callno) \ > +({ \ > + struct { unsigned char c[2]; } *p = (void *) ((state)->rip - 2); \ > + int result; \ > + if (_hurdsig_catch_memory_fault (p)) \ > + return 0; \ > + if (result = p->c[0] == 0xf && p->c[1] == 0x5) \ > + /* The PC appears to be just after a `syscall' instruction. \ > + This is a system call in progress; %rax holds the call number. */ \ > + *(callno) = (state)->rax; \ > + _hurdsig_end_catch_fault (); \ > + result; \ > +}) > + > + > +/* This cannot be an inline function because it calls setjmp. */ > +#define MSG_EXAMINE(state, msgid, rcvname, send_name, opt, tmout) \ > +({ \ > + int ret = 0; \ > + const struct machine_thread_state *s = (state); \ > + const mach_msg_header_t *msg = (const void *) s->rdi; \ > + *(rcvname) = s->r8; \ > + *(opt) = s->rsi; \ > + *(tmout) = s->r9; \ > + if (msg == 0) \ > + { \ > + *(send_name) = MACH_PORT_NULL; \ > + *(msgid) = 0; \ > + } \ > + else \ > + { \ > + ret = _hurdsig_catch_memory_fault (msg) ? -1 : 0; \ > + if (ret == 0) \ > + { \ > + *(send_name) = msg->msgh_remote_port; \ > + *(msgid) = msg->msgh_id; \ > + _hurdsig_end_catch_fault (); \ > + } \ > + } \ > + ret; \ > +}) > -- > 2.39.2 > -- Samuel --- Pour une évaluation indépendante, transparente et rigoureuse ! Je soutiens la Commission d'Évaluation de l'Inria.