From: Ammar Faizi <[email protected]>
To: Jens Axboe <[email protected]>
Cc: Ammar Faizi <[email protected]>,
io-uring Mailing List <[email protected]>,
GNU/Weeb Mailing List <[email protected]>,
Pavel Begunkov <[email protected]>,
David Laight <[email protected]>,
Alviro Iskandar Setiawan <[email protected]>
Subject: [PATCH liburing 3/3] arch/x86/syscall: Add x86 32-bit native syscall support
Date: Fri, 15 Apr 2022 05:41:40 +0700 [thread overview]
Message-ID: <[email protected]> (raw)
In-Reply-To: <[email protected]>
Create __do_syscall{0..6} macros for x86 32-bit. Unlike x86-64, only
use these macros when CONFIG_NOLIBC is enabled for a 32-bit build. The
reason is that the libc syscall wrapper can do better in 32-bit.
libc syscall wrapper can dispatch the best syscall instruction that the
environment is supported, there are at least two variants of syscall
instruction for x86 32-bit, they are: `int $0x80` and `sysenter`. The
`int $0x80` instruction is always available, but `sysenter` is not, it
relies on VDSO. liburing always uses `int $0x80` for syscall if it's
compiled with CONFIG_NOLIBC, otherwise, it uses whatever the libc
provides.
Extra notes for __do_syscall6() macro:
On i386, the 6th argument of syscall goes in %ebp. However, both Clang
and GCC cannot use %ebp in the clobber list and the "r" constraint
without using -fomit-frame-pointer. To make it always available for any
kind of compilation, the below workaround is implemented:
1) Push the 6-th argument.
2) Push %ebp.
3) Load the 6-th argument from 4(%esp) to %ebp.
4) Do the syscall (int $0x80).
5) Pop %ebp (restore the old value of %ebp).
6) Add %esp by 4 (undo the stack pointer).
WARNING:
Don't use register variables for __do_syscall6(), there is a known
GCC bug that results in an endless loop.
BugLink: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105032
Link: https://lore.kernel.org/lkml/[email protected]
Suggested-by: David Laight <[email protected]>
Co-authored-by: Alviro Iskandar Setiawan <[email protected]>
Signed-off-by: Alviro Iskandar Setiawan <[email protected]>
Signed-off-by: Ammar Faizi <[email protected]>
---
src/arch/x86/syscall.h | 150 ++++++++++++++++++++++++++++++++++++++++-
1 file changed, 148 insertions(+), 2 deletions(-)
diff --git a/src/arch/x86/syscall.h b/src/arch/x86/syscall.h
index 89a68f6..8cd24dd 100644
--- a/src/arch/x86/syscall.h
+++ b/src/arch/x86/syscall.h
@@ -151,10 +151,156 @@
* TODO: Add x86 (32-bit) nolibc support.
*/
#ifdef CONFIG_NOLIBC
- #error "x86 (32-bit) is currently not supported for nolibc builds"
-#endif
+/**
+ * Note for syscall registers usage (x86, 32-bit):
+ * - %eax is the syscall number.
+ * - %eax is also the return value.
+ * - %ebx is the 1st argument.
+ * - %ecx is the 2nd argument.
+ * - %edx is the 3rd argument.
+ * - %esi is the 4th argument.
+ * - %edi is the 5th argument.
+ * - %ebp is the 6th argument.
+ */
+
+#define __do_syscall0(NUM) ({ \
+ intptr_t eax; \
+ \
+ __asm__ volatile( \
+ "int $0x80" \
+ : "=a"(eax) /* %eax */ \
+ : "a"(NUM) /* %eax */ \
+ : "memory" \
+ ); \
+ eax; \
+})
+
+#define __do_syscall1(NUM, ARG1) ({ \
+ intptr_t eax; \
+ \
+ __asm__ volatile( \
+ "int $0x80" \
+ : "=a"(eax) /* %eax */ \
+ : "a"(NUM), /* %eax */ \
+ "b"((ARG1)) /* %ebx */ \
+ : "memory" \
+ ); \
+ eax; \
+})
+
+#define __do_syscall2(NUM, ARG1, ARG2) ({ \
+ intptr_t eax; \
+ \
+ __asm__ volatile( \
+ "int $0x80" \
+ : "=a" (eax) /* %eax */ \
+ : "a"(NUM), /* %eax */ \
+ "b"((ARG1)), /* %ebx */ \
+ "c"((ARG2)) /* %ecx */ \
+ : "memory" \
+ ); \
+ eax; \
+})
+
+#define __do_syscall3(NUM, ARG1, ARG2, ARG3) ({ \
+ intptr_t eax; \
+ \
+ __asm__ volatile( \
+ "int $0x80" \
+ : "=a" (eax) /* %eax */ \
+ : "a"(NUM), /* %eax */ \
+ "b"((ARG1)), /* %ebx */ \
+ "c"((ARG2)), /* %ecx */ \
+ "d"((ARG3)) /* %edx */ \
+ : "memory" \
+ ); \
+ eax; \
+})
+
+#define __do_syscall4(NUM, ARG1, ARG2, ARG3, ARG4) ({ \
+ intptr_t eax; \
+ \
+ __asm__ volatile( \
+ "int $0x80" \
+ : "=a" (eax) /* %eax */ \
+ : "a"(NUM), /* %eax */ \
+ "b"((ARG1)), /* %ebx */ \
+ "c"((ARG2)), /* %ecx */ \
+ "d"((ARG3)), /* %edx */ \
+ "S"((ARG4)) /* %esi */ \
+ : "memory" \
+ ); \
+ eax; \
+})
+
+#define __do_syscall5(NUM, ARG1, ARG2, ARG3, ARG4, ARG5) ({ \
+ intptr_t eax; \
+ \
+ __asm__ volatile( \
+ "int $0x80" \
+ : "=a" (eax) /* %eax */ \
+ : "a"(NUM), /* %eax */ \
+ "b"((ARG1)), /* %ebx */ \
+ "c"((ARG2)), /* %ecx */ \
+ "d"((ARG3)), /* %edx */ \
+ "S"((ARG4)), /* %esi */ \
+ "D"((ARG5)) /* %edi */ \
+ : "memory" \
+ ); \
+ eax; \
+})
+
+
+/*
+ * On i386, the 6th argument of syscall goes in %ebp. However, both Clang
+ * and GCC cannot use %ebp in the clobber list and in the "r" constraint
+ * without using -fomit-frame-pointer. To make it always available for
+ * any kind of compilation, the below workaround is implemented:
+ *
+ * 1) Push the 6-th argument.
+ * 2) Push %ebp.
+ * 3) Load the 6-th argument from 4(%esp) to %ebp.
+ * 4) Do the syscall (int $0x80).
+ * 5) Pop %ebp (restore the old value of %ebp).
+ * 6) Add %esp by 4 (undo the stack pointer).
+ *
+ * WARNING:
+ * Don't use register variables for __do_syscall6(), there is a known
+ * GCC bug that results in an endless loop.
+ *
+ * BugLink: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=105032
+ *
+ */
+#define __do_syscall6(NUM, ARG1, ARG2, ARG3, ARG4, ARG5, ARG6) ({ \
+ intptr_t eax = (intptr_t)(NUM); \
+ intptr_t arg6 = (intptr_t)(ARG6); /* Always in memory */ \
+ __asm__ volatile ( \
+ "pushl %[_arg6]\n\t" \
+ "pushl %%ebp\n\t" \
+ "movl 4(%%esp),%%ebp\n\t" \
+ "int $0x80\n\t" \
+ "popl %%ebp\n\t" \
+ "addl $4,%%esp" \
+ : "+a"(eax) /* %eax */ \
+ : "b"(ARG1), /* %ebx */ \
+ "c"(ARG2), /* %ecx */ \
+ "d"(ARG3), /* %edx */ \
+ "S"(ARG4), /* %esi */ \
+ "D"(ARG5), /* %edi */ \
+ [_arg6]"m"(arg6) /* memory */ \
+ : "memory", "cc" \
+ ); \
+ eax; \
+})
+
+#include "../syscall-defs.h"
+
+#else /* #ifdef CONFIG_NOLIBC */
+
#include "../generic/syscall.h"
+#endif /* #ifdef CONFIG_NOLIBC */
+
#endif /* #if defined(__x86_64__) */
#endif /* #ifndef LIBURING_ARCH_X86_SYSCALL_H */
--
Ammar Faizi
next prev parent reply other threads:[~2022-04-14 22:42 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-04-14 22:41 [PATCH liburing 0/3] Add x86 32-bit support for the nolibc build Ammar Faizi
2022-04-14 22:41 ` [PATCH liburing 1/3] arch/syscall-defs: Use `__NR_mmap2` instead of `__NR_mmap` for x86 32-bit Ammar Faizi
2022-04-14 22:41 ` [PATCH liburing 2/3] arch/x86/lib: Provide `get_page_size()` function " Ammar Faizi
2022-04-14 22:41 ` Ammar Faizi [this message]
2022-04-18 2:01 ` [PATCH liburing 0/3] Add x86 32-bit support for the nolibc build Jens Axboe
2022-04-18 14:14 ` Ammar Faizi
2022-04-18 15:23 ` Jens Axboe
2022-04-18 15:24 ` Jens Axboe
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 \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
[email protected] \
/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