From: mengqinggang <mengqinggang@loongson.cn>
To: binutils@sourceware.org
Cc: xuchenghua@loongson.cn, chenglulu@loongson.cn,
liuzhensong@loongson.cn, xry111@xry111.site, i.swmail@xen0n.name,
maskray@google.com, hejinyang@loongson.cn,
mengqinggang <mengqinggang@loongson.cn>
Subject: [PATCH 1/2] LoongArch: Fix instruction immediate bug caused by sign-extend
Date: Mon, 17 Jul 2023 16:22:28 +0800 [thread overview]
Message-ID: <20230717082229.2149099-1-mengqinggang@loongson.cn> (raw)
For extreme code mode, the instruction sequences is
pcalau12i $t0, hi20
addi.d $t1, $zero, lo12
lu32i.d $t1, lo20
lu52i.d $t1, hi12
add.d $t1, $t0, $t1
If lo12 > 0x7ff, hi20 need to add 0x1, lo20 need to sub 0x1.
If hi20 > 0x7ffff, lo20 need to add 0x1.
bfd/ChangeLog:
* elfnn-loongarch.c (RELOCATE_CALC_PC32_HI20): Redefined.
(RELOCATE_CALC_PC64_HI32): Redefined.
---
bfd/elfnn-loongarch.c | 59 +++++++++++++++++++++++++++++++++++--------
1 file changed, 49 insertions(+), 10 deletions(-)
diff --git a/bfd/elfnn-loongarch.c b/bfd/elfnn-loongarch.c
index d3d8419d80b..e9c408b2dff 100644
--- a/bfd/elfnn-loongarch.c
+++ b/bfd/elfnn-loongarch.c
@@ -2284,26 +2284,65 @@ loongarch_reloc_is_fatal (struct bfd_link_info *info,
return fatal;
}
+/* If lo12 immediate > 0x7ff, because sign-extend caused by addi.d/ld.d,
+ hi20 immediate need to add 0x1.
+ For example: pc 0x120000000, symbol 0x120000812
+ lo12 immediate is 0x812, 0x120000812 & 0xfff = 0x812
+ hi20 immediate is 1, because lo12 imm > 0x7ff, symbol need to add 0x1000
+ (((0x120000812 + 0x1000) & ~0xfff) - (0x120000000 & ~0xfff)) >> 12 = 0x1
+
+ At run:
+ pcalau12i $t0, hi20 (0x1)
+ $t0 = 0x120000000 + (0x1 << 12) = 0x120001000
+ addi.d $t0, $t0, lo12 (0x812)
+ $t0 = 0x120001000 + 0xfffffffffffff812 (-(0x1000 - 0x812) = -0x7ee)
+ = 0x120001000 - 0x7ee (0x1000 - 0x7ee = 0x812)
+ = 0x120000812
+ Without hi20 add 0x1000, the result 0x120000000 - 0x7ee = 0x11ffff812 is
+ error.
+ 0x1000 + sign-extend-to64(0x8xx) = 0x8xx. */
#define RELOCATE_CALC_PC32_HI20(relocation, pc) \
({ \
bfd_vma __lo = (relocation) & ((bfd_vma)0xfff); \
- pc = pc & (~(bfd_vma)0xfff); \
+ relocation = (relocation & ~(bfd_vma)0xfff) \
+ - (pc & ~(bfd_vma)0xfff); \
if (__lo > 0x7ff) \
- { \
relocation += 0x1000; \
- } \
- relocation &= ~(bfd_vma)0xfff; \
- relocation -= pc; \
})
+/* For example: pc is 0x11000010000100, symbol is 0x1812348ffff812
+ offset = (0x1812348ffff812 & ~0xfff) - (0x11000010000100 & ~0xfff)
+ = 0x712347ffff000
+ lo12: 0x1812348ffff812 & 0xfff = 0x812
+ hi20: 0x7ffff + 0x1(lo12 > 0x7ff) = 0x80000
+ lo20: 0x71234 - 0x1(lo12 > 0x7ff) + 0x1(hi20 > 0x7ffff)
+ hi12: 0x0
+
+ pcalau12i $t1, hi20 (0x80000)
+ $t1 = 0x11000010000100 + sign-extend(0x80000 << 12)
+ = 0x11000010000100 + 0xffffffff80000000
+ = 0x10ffff90000000
+ addi.d $t0, $zero, lo12 (0x812)
+ $t0 = 0xfffffffffffff812 (if lo12 > 0x7ff, because sign-extend,
+ lo20 need to sub 0x1)
+ lu32i.d $t0, lo12 (0x71234)
+ $t0 = {0x71234, 0xfffff812}
+ = 0x71234fffff812
+ lu52i.d $t0, hi12 (0x0)
+ $t0 = {0x0, 0x71234fffff812}
+ = 0x71234fffff812
+ add.d $t1, $t1, $t0
+ $t1 = 0x10ffff90000000 + 0x71234fffff812
+ = 0x1812348ffff812. */
#define RELOCATE_CALC_PC64_HI32(relocation, pc) \
({ \
- bfd_vma __lo = (relocation) & ((bfd_vma)0xfff); \
+ bfd_vma __lo = (relocation & (bfd_vma)0xfff); \
+ relocation = (relocation & ~(bfd_vma)0xfff) \
+ - (pc & ~(bfd_vma)0xfff); \
if (__lo > 0x7ff) \
- { \
- relocation -= 0x100000000; \
- } \
- relocation -= (pc & ~(bfd_vma)0xffffffff); \
+ relocation += (0x1000 - 0x100000000); \
+ if (relocation & 0x80000000) \
+ relocation += 0x100000000; \
})
static int
--
2.36.0
next reply other threads:[~2023-07-17 8:22 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-07-17 8:22 mengqinggang [this message]
2023-07-17 8:22 ` [PATCH 2/2] LoongArch: Fix immediate overflow check bug mengqinggang
2023-07-21 3:56 ` mengqinggang
2023-07-22 7:47 ` Xi Ruoyao
2023-07-22 7:52 ` Xi Ruoyao
2023-07-22 8:23 ` Jinyang He
2023-07-24 9:14 ` Nick Clifton
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=20230717082229.2149099-1-mengqinggang@loongson.cn \
--to=mengqinggang@loongson.cn \
--cc=binutils@sourceware.org \
--cc=chenglulu@loongson.cn \
--cc=hejinyang@loongson.cn \
--cc=i.swmail@xen0n.name \
--cc=liuzhensong@loongson.cn \
--cc=maskray@google.com \
--cc=xry111@xry111.site \
--cc=xuchenghua@loongson.cn \
/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).