From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
To: Palmer Dabbelt <palmer@rivosinc.com>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH 3/5] riscv: Add the clone3 wrapper
Date: Mon, 29 May 2023 17:38:50 -0300 [thread overview]
Message-ID: <9864c8ad-c2ef-56fc-796d-13f20de0e94f@linaro.org> (raw)
In-Reply-To: <mhng-4a9d49cc-5719-4d1e-9cc4-f6db86c9377d@palmer-ri-x1c9>
On 29/05/23 11:04, Palmer Dabbelt wrote:
> On Fri, 03 Feb 2023 09:12:35 PST (-0800), adhemerval.zanella@linaro.org wrote:
>> It follows the internal signature:
>>
>> extern int clone3 (struct clone_args *__cl_args, size_t __size,
>> int (*__func) (void *__arg), void *__arg);
>>
>> Checked on riscv64-linux-gnu-rv64imafdc-lp64d.
>
> Due to a bug we only had clone3 on rv64 until 59a4e0d5511b ("RISC-V: Include clone3() on rv32"), which landed in 5.15 and was then backported all the way to 5.4.
>
>> ---
>> sysdeps/unix/sysv/linux/riscv/clone3.S | 80 ++++++++++++++++++++++++++
>> sysdeps/unix/sysv/linux/riscv/sysdep.h | 1 +
>> 2 files changed, 81 insertions(+)
>> create mode 100644 sysdeps/unix/sysv/linux/riscv/clone3.S
>>
>> diff --git a/sysdeps/unix/sysv/linux/riscv/clone3.S b/sysdeps/unix/sysv/linux/riscv/clone3.S
>> new file mode 100644
>> index 0000000000..ee5780ee2f
>> --- /dev/null
>> +++ b/sysdeps/unix/sysv/linux/riscv/clone3.S
>> @@ -0,0 +1,80 @@
>> +/* The clone3 syscall wrapper. Linux/RISC-V version.
>> + Copyright (C) 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
>> + <https://www.gnu.org/licenses/>. */
>> +
>> +#include <asm/errno.h>
>> +#include <sys/asm.h>
>> +#include <sysdep.h>
>> +
>> +/* The userland implementation is:
>> + int clone3 (struct clone_args *cl_args, size_t size,
>> + int (*func)(void *arg), void *arg);
>> +
>> + the kernel entry is:
>> + int clone3 (struct clone_args *cl_args, size_t size);
>> +
>> + The parameters are passed in registers from userland:
>> + a0: cl_args
>> + a1: size
>> + a2: func
>> + a3: arg */
>> +
>> + .text
>> +ENTRY(__clone3)
>> + /* Sanity check args. */
>> + beqz a0, L(invalid) /* No NULL cl_args pointer. */
>> + beqz a2, L(invalid) /* No NULL function pointer. */
>> +
>> + /* Do the system call, the kernel expects:
>> + a7: system call number
>> + a0: cl_args
>> + a1: size */
>> + li a7, __NR_clone3
>
> IIUC this only builds with kernel headers 5.15 or newer on rv32. I'm not sure if that's a problem: unless I'm missing something we support building for those systems now, but rv32 is pretty esoteric so I'm not sure anyone cares -- maybe someone uses 5.15 on rv32 (some rv64 hardware is still being released with 5.10-based vendor kernels), but 5.4 is pretty old for us.
The __NR_XXX macros are now provided by arch-syscall.h, which in turn is generated
by the kernel header. And we already have clone3 for rv32 since 3387c40a8bb.
>
>> + scall
>> +
>> + bltz a0, L(error)
>> + beqz a0, L(thread_start)
>> +
>> + ret
>> +
>> +L(invalid):
>> + li a0, -EINVAL
>> +L(error):
>> + tail __syscall_error
>> +END (__clone3)
>> +
>> +ENTRY(__thread_start_clone3)
>> +L(thread_start):
>> + /* Terminate call stack by noting ra is undefined. Use a dummy
>> + .cfi_label to force starting the FDE. */
>> + .cfi_label .Ldummy
>> + cfi_undefined (ra)
>> +
>> + /* Restore the arg for user's function and call the user's
>> + function. */
>> + mv a1, a2 /* Function pointer. */
>
> I think we don't need that mv, we can just call a2 directly? Unless I'm missing some reason the thread main needs to see a pointer to itself?
We can, I will change it.
>
>> + mv a0, a3 /* Argument pointer. */
>> + jalr a1
>> +
>> + /* Call exit with the function's return value. */
>> + li a7, __NR_exit
>> + scall
>> +END(__thread_start_clone3)
>> +
>> +libc_hidden_def (__clone3)
>> +weak_alias (__clone3, clone3)
>> diff --git a/sysdeps/unix/sysv/linux/riscv/sysdep.h b/sysdeps/unix/sysv/linux/riscv/sysdep.h
>> index 4af5fe5dbc..e96a930409 100644
>> --- a/sysdeps/unix/sysv/linux/riscv/sysdep.h
>> +++ b/sysdeps/unix/sysv/linux/riscv/sysdep.h
>> @@ -151,6 +151,7 @@
>>
>> /* RV32 does not support the gettime VDSO syscalls. */
>> # endif
>> +# define HAVE_CLONE3_WRAPPER 1
>>
>> /* List of system calls which are supported as vsyscalls (for RV32 and
>> RV64). */
>
> That's just some minor comments though, so
>
> Reviewed-by: Palmer Dabbelt <palmer@rivosinc.com>
Thanks.
>
> Thanks! (and sorry I missed it for a bit)
next prev parent reply other threads:[~2023-05-29 20:38 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-02-03 17:12 [PATCH 0/5] Add clone3 support for multiple architectures Adhemerval Zanella
2023-02-03 17:12 ` [PATCH 1/5] powerpc64: Add the clone3 wrapper Adhemerval Zanella
2023-02-03 21:13 ` Paul E Murphy
2023-02-03 17:12 ` [PATCH 2/5] s390x: " Adhemerval Zanella
2023-05-25 17:36 ` Adhemerval Zanella Netto
2023-02-03 17:12 ` [PATCH 3/5] riscv: " Adhemerval Zanella
2023-05-25 17:36 ` Adhemerval Zanella Netto
2023-05-29 14:04 ` Palmer Dabbelt
2023-05-29 20:38 ` Adhemerval Zanella Netto [this message]
2023-02-03 17:12 ` [PATCH 4/5] arm: " Adhemerval Zanella
2023-05-25 17:36 ` Adhemerval Zanella Netto
2023-02-03 17:12 ` [PATCH 5/5] mips: " Adhemerval Zanella
2023-05-25 17:36 ` Adhemerval Zanella Netto
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=9864c8ad-c2ef-56fc-796d-13f20de0e94f@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=libc-alpha@sourceware.org \
--cc=palmer@rivosinc.com \
/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).