From: Bo Gan <ganboing@gmail.com>
To: opensbi@lists.infradead.org, wangruikang@iscas.ac.cn,
dramforever@live.com, andrew.jones@oss.qualcomm.com
Cc: cleger@rivosinc.com, pjw@kernel.org, asrinivasan@oss.tenstorrent.com
Subject: [PATCH v2 2/4] lib: sbi: Rework and split sbi_misaligned(_v)_tinst_fixup
Date: Mon, 8 Jun 2026 23:00:22 -0700 [thread overview]
Message-ID: <20260609060024.706-3-ganboing@gmail.com> (raw)
In-Reply-To: <20260609060024.706-1-ganboing@gmail.com>
The load/store address offset between the uptrap and the orig_trap
can be derived by orig_trap->tval - uptrap->tval, thus refactor
the function prototype for simplicity.
For vector load, sbi_misaligned_v_tinst_fixup is introduced. There's
no transformed instruction for vector load/store, so null out tinst
if the fault is not a guest-page fault.
Signed-off-by: Bo Gan <ganboing@gmail.com>
---
include/sbi/sbi_trap_ldst.h | 3 ---
lib/sbi/sbi_trap_ldst.c | 46 +++++++++++++++++++++++++++----------
lib/sbi/sbi_trap_v_ldst.c | 31 ++++++++++++++++++++-----
3 files changed, 59 insertions(+), 21 deletions(-)
diff --git a/include/sbi/sbi_trap_ldst.h b/include/sbi/sbi_trap_ldst.h
index 33c348c5..30228d24 100644
--- a/include/sbi/sbi_trap_ldst.h
+++ b/include/sbi/sbi_trap_ldst.h
@@ -28,9 +28,6 @@ int sbi_load_access_handler(struct sbi_trap_context *tcntx);
int sbi_store_access_handler(struct sbi_trap_context *tcntx);
-ulong sbi_misaligned_tinst_fixup(ulong orig_tinst, ulong new_tinst,
- ulong addr_offset);
-
int sbi_misaligned_v_ld_emulator(ulong insn,
struct sbi_trap_context *tcntx);
diff --git a/lib/sbi/sbi_trap_ldst.c b/lib/sbi/sbi_trap_ldst.c
index c1392251..5f7de662 100644
--- a/lib/sbi/sbi_trap_ldst.c
+++ b/lib/sbi/sbi_trap_ldst.c
@@ -32,16 +32,40 @@ typedef int (*sbi_trap_st_emulator)(ulong insn, int wlen, ulong waddr,
union sbi_ldst_data in_val,
struct sbi_trap_context *tcntx);
-ulong sbi_misaligned_tinst_fixup(ulong orig_tinst, ulong new_tinst,
- ulong addr_offset)
+/**
+ * Handling of misaligned fault is done by a collection of smaller, but
+ * aligned load/store(s). Another fault (load/store, page fault...) can
+ * arise from any of them, then the handling gets aborted. We must fixup
+ * the tinst to pretend the fault was rised from the original insn.
+ * Specifically, fixup the offset field using the tval diff between the
+ * new trap and the original one (if required).
+ */
+static inline void sbi_misaligned_tinst_fixup(
+ const struct sbi_trap_info *orig_trap,
+ struct sbi_trap_info *uptrap)
{
- if (new_tinst == INSN_PSEUDO_VS_LOAD ||
- new_tinst == INSN_PSEUDO_VS_STORE)
- return new_tinst;
- else if (orig_tinst == 0)
- return 0UL;
+ ulong offset = uptrap->tval - orig_trap->tval;
+
+ /*
+ * The function is called in code path for handling a scalar
+ * load/store misaligned fault, thus the new uptrap can't have
+ * custom value of tinst
+ */
+ if (uptrap->tinst == INSN_PSEUDO_VS_LOAD ||
+ uptrap->tinst == INSN_PSEUDO_VS_STORE)
+ /* Use uptrap as-is for guest-page faults */
+ return;
+ /*
+ * Only fixup if orig tinst is valid. Otherwise, discard the
+ * new tinst to be on the safe side. Never use new tinst as-is!
+ * It's load/store width surely mismatches the original width.
+ * For vector, discard it regardless. It doesn't make sense to
+ * have a transformed tinst
+ */
+ else if (orig_trap->tinst == 0)
+ uptrap->tinst = 0;
else
- return orig_tinst | (addr_offset << SH_RS1);
+ uptrap->tinst = orig_trap->tinst | (offset << SH_RS1);
}
static inline bool sbi_trap_tinst_valid(ulong tinst)
@@ -464,8 +488,7 @@ static int sbi_misaligned_ld_emulator(ulong insn, int rlen, ulong addr,
out_val->data_bytes[i] =
sbi_load_u8((void *)(addr + i), &uptrap);
if (uptrap.cause) {
- uptrap.tinst = sbi_misaligned_tinst_fixup(
- orig_trap->tinst, uptrap.tinst, i);
+ sbi_misaligned_tinst_fixup(orig_trap, &uptrap);
return sbi_trap_redirect(regs, &uptrap);
}
}
@@ -501,8 +524,7 @@ static int sbi_misaligned_st_emulator(ulong insn, int wlen, ulong addr,
sbi_store_u8((void *)(addr + i),
in_val.data_bytes[i], &uptrap);
if (uptrap.cause) {
- uptrap.tinst = sbi_misaligned_tinst_fixup(
- orig_trap->tinst, uptrap.tinst, i);
+ sbi_misaligned_tinst_fixup(orig_trap, &uptrap);
return sbi_trap_redirect(regs, &uptrap);
}
}
diff --git a/lib/sbi/sbi_trap_v_ldst.c b/lib/sbi/sbi_trap_v_ldst.c
index 7d2e1409..e16e3def 100644
--- a/lib/sbi/sbi_trap_v_ldst.c
+++ b/lib/sbi/sbi_trap_v_ldst.c
@@ -138,9 +138,31 @@ static inline void vsetvl(ulong vl, ulong vtype)
:: "r" (vl), "r" (vtype));
}
+/**
+ * Handling of misaligned fault is done by a collection of smaller, but
+ * aligned load/store(s). Another fault (load/store, page fault...) can
+ * arise from any of them, then the handling gets aborted. We must fixup
+ * the tinst to pretend the fault was rised from the original insn. For
+ * vector insn, simply null out tinst if it's not a guest-page fault, as
+ * there's no transformed insn for vector load/store
+ */
+static inline void sbi_misaligned_v_tinst_fixup(struct sbi_trap_info *uptrap)
+{
+ /*
+ * The function is called in code path for handling a vector
+ * load/store misaligned fault, thus the new uptrap can't have
+ * custom value of tinst
+ */
+ if (uptrap->tinst == INSN_PSEUDO_VS_LOAD ||
+ uptrap->tinst == INSN_PSEUDO_VS_STORE)
+ /* Use uptrap as-is for guest-page faults */
+ return;
+
+ uptrap->tinst = 0;
+}
+
int sbi_misaligned_v_ld_emulator(ulong insn, struct sbi_trap_context *tcntx)
{
- const struct sbi_trap_info *orig_trap = &tcntx->trap;
struct sbi_trap_regs *regs = &tcntx->regs;
struct sbi_trap_info uptrap;
ulong vl = csr_read(CSR_VL);
@@ -218,8 +240,7 @@ int sbi_misaligned_v_ld_emulator(ulong insn, struct sbi_trap_context *tcntx)
break;
}
vsetvl(vl, vtype);
- uptrap.tinst = sbi_misaligned_tinst_fixup(
- orig_trap->tinst, uptrap.tinst, i);
+ sbi_misaligned_v_tinst_fixup(&uptrap);
return sbi_trap_redirect(regs, &uptrap);
}
}
@@ -240,7 +261,6 @@ int sbi_misaligned_v_ld_emulator(ulong insn, struct sbi_trap_context *tcntx)
int sbi_misaligned_v_st_emulator(ulong insn, struct sbi_trap_context *tcntx)
{
- const struct sbi_trap_info *orig_trap = &tcntx->trap;
struct sbi_trap_regs *regs = &tcntx->regs;
struct sbi_trap_info uptrap;
ulong vl = csr_read(CSR_VL);
@@ -317,8 +337,7 @@ int sbi_misaligned_v_st_emulator(ulong insn, struct sbi_trap_context *tcntx)
bytes[seg * len + i], &uptrap);
if (uptrap.cause) {
vsetvl(vl, vtype);
- uptrap.tinst = sbi_misaligned_tinst_fixup(
- orig_trap->tinst, uptrap.tinst, i);
+ sbi_misaligned_v_tinst_fixup(&uptrap);
return sbi_trap_redirect(regs, &uptrap);
}
}
--
2.34.1
--
opensbi mailing list
opensbi@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
next prev parent reply other threads:[~2026-06-09 6:02 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-06-09 6:00 [PATCH v2 0/4] Fixes for vector misaligned load/store handlers Bo Gan
2026-06-09 6:00 ` [PATCH v2 1/4] lib: sbi: cosmetic changes to reduce indentation Bo Gan
2026-06-09 6:00 ` Bo Gan [this message]
2026-06-09 6:00 ` [PATCH v2 3/4] lib: sbi: Add variable-length unprivilege access functions Bo Gan
2026-06-09 6:00 ` [PATCH v2 4/4] lib: sbi: Rework misaligned vector load/store Bo Gan
2026-06-09 22:02 ` [PATCH v2 0/4] Fixes for vector misaligned load/store handlers Anirudh Srinivasan
2026-06-09 23:54 ` Bo Gan
2026-06-09 23:59 ` Anirudh Srinivasan
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=20260609060024.706-3-ganboing@gmail.com \
--to=ganboing@gmail.com \
--cc=andrew.jones@oss.qualcomm.com \
--cc=asrinivasan@oss.tenstorrent.com \
--cc=cleger@rivosinc.com \
--cc=dramforever@live.com \
--cc=opensbi@lists.infradead.org \
--cc=pjw@kernel.org \
--cc=wangruikang@iscas.ac.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