Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Tue, 27 Apr 2021 19:07:51 -0700
User-Agent: K-9 Mail for Android
In-Reply-To: <202104271857.4708C9A25@keescook>
References: <202104271619.0DBE456@keescook> <78327AF2-575F-415E-98D4-6ECDFE311D0F@amacapital.net> <202104271857.4708C9A25@keescook>
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
Subject: Re: pt_regs->ax == -ENOSYS
To:     Kees Cook <keescook@chromium.org>,
        Andy Lutomirski <luto@amacapital.net>
CC:     Linus Torvalds <torvalds@linux-foundation.org>,
        Ingo Molnar <mingo@kernel.org>,
        Thomas Gleixner <tglx@linutronix.de>,
        Andrew Lutomirski <luto@kernel.org>,
        Borislav Petkov <bp@alien8.de>, linux-kernel@vger.kernel.org,
        oleg@redhat.com, Will Drewry <wad@chromium.org>
From:   "H. Peter Anvin" <hpa@zytor.com>
Message-ID: <C9135D5F-AEB4-4093-A54B-8AD99AD570C4@zytor.com>
Precedence: bulk

Earlier in the thread the suggestion was to have (int)pt_regs->orig_ax < 0 =
indicate a nonsyscall=2E

On April 27, 2021 7:05:56 PM PDT, Kees Cook <keescook@chromium=2Eorg> wrot=
e:
>On Tue, Apr 27, 2021 at 04:51:06PM -0700, Andy Lutomirski wrote:
>> Fortunately there is not, and never will be, a syscall -1=2E  But I
>> agree that calling max syscall + 1 should behave identically to
>calling
>> a nonexistent syscall in the middle of the table=2E
>
>If that happens, we have to separate the meaning of -1L from ptrace,
>seccomp, etc=2E (i=2Ee=2E we can't just add an "else { result =3D -ENOSYS=
; }"
>to
>the syscall table dispatching code, since that'll overwrite any written
>return value when the syscall is meant to be skipped with a specific
>return value set by ptrace/seccomp=2E
>
>syscall_trace_enter() will currently return either -1 or the
>syscall=2E Which means someone making a "syscall -1" will get the skip
>semantics currently (though the preloaded -ENOSYS results in the
>"expected" outcome)=2E
>
>arm64 recently had to untangle this too:
>
>15956689a0e6 arm64: compat: Ensure upper 32 bits of x0 are zero on
>syscall return
>59ee987ea47c arm64: ptrace: Add a comment describing our syscall
>entry/exit trap ABI
>139dbe5d8ed3 arm64: syscall: Expand the comment about ptrace and
>syscall(-1)
>d83ee6e3e75d arm64: ptrace: Use NO_SYSCALL instead of -1 in
>syscall_trace_enter()

--=20
Sent from my Android device with K-9 Mail=2E Please excuse my brevity=2E