From: Zong Li <zong.li@sifive.com>
To: opensbi@lists.infradead.org, anup@brainfault.org
Cc: Zong Li <zong.li@sifive.com>
Subject: [PATCH 2/3] firmware: fw_base.S: override a0, a1 and a2 from the firmware header
Date: Mon, 31 Aug 2026 19:22:10 -0700 [thread overview]
Message-ID: <20260901022211.2619560-3-zong.li@sifive.com> (raw)
In-Reply-To: <20260901022211.2619560-1-zong.li@sifive.com>
Wire up the flags word and the three override values in the OpenSBI
firmware header. When the previous booting stage sets one of the
FW_HEADER_FLAGS_OVERRIDE_A[012] bits, the cold boot path replaces the
matching register with the value stored in the header before doing
anything else with the boot arguments.
This lets a previous booting stage which cannot pass the boot arguments
in registers hand them over by patching a few words in the firmware
image instead. For example, a booting stage running on a dedicated boot
processor can load the OpenSBI image, patch the header with the hart id,
the DTB address and, for FW_DYNAMIC, the address of a struct
fw_dynamic_info it built somewhere in DRAM, and then release the hart
that runs OpenSBI, without ever being able to set up that hart's
registers itself.
The flags word is zero in a freshly built image, so nothing is
overridden and every existing booting stage keeps passing a0, a1 and a2
in registers as before.
Suggested-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Zong Li <zong.li@sifive.com>
---
firmware/fw_base.S | 32 ++++++++++++++++++++++++++++++++
1 file changed, 32 insertions(+)
diff --git a/firmware/fw_base.S b/firmware/fw_base.S
index dd2adb8a..bf458ba0 100644
--- a/firmware/fw_base.S
+++ b/firmware/fw_base.S
@@ -114,6 +114,38 @@ _fw_header_override_a2:
/* Reserved, pads the header up to FW_HEADER_SIZE bytes */
.fill (FW_HEADER_SIZE - FW_HEADER_RESERVED_OFFSET), 1, 0
_start_real:
+ /*
+ * Override a0, a1 and a2 with the values from the firmware header.
+ *
+ * This is for a previous booting stage which cannot pass the boot
+ * arguments in registers, but can patch the firmware image before
+ * jumping to it. Nothing is overridden when the previous booting
+ * stage left the flags word alone, so the registers passed by all
+ * existing booting stages are used as-is.
+ *
+ * This runs before relocation, so only PC-relative addressing is
+ * used. t0, t1 and t2 are free to use here: the boot arguments live
+ * in a0-a4 and every other register is reset further down the cold
+ * boot path.
+ */
+ lla t0, _fw_header_flags
+ lw t0, (t0)
+ andi t1, t0, FW_HEADER_FLAGS_OVERRIDE_A0
+ beqz t1, _skip_override_a0
+ lla t2, _fw_header_override_a0
+ REG_L a0, (t2)
+_skip_override_a0:
+ andi t1, t0, FW_HEADER_FLAGS_OVERRIDE_A1
+ beqz t1, _skip_override_a1
+ lla t2, _fw_header_override_a1
+ REG_L a1, (t2)
+_skip_override_a1:
+ andi t1, t0, FW_HEADER_FLAGS_OVERRIDE_A2
+ beqz t1, _skip_override_a2
+ lla t2, _fw_header_override_a2
+ REG_L a2, (t2)
+_skip_override_a2:
+
/* Find preferred boot HART id */
MOV_3R s0, a0, s1, a1, s2, a2
call fw_boot_hart
--
2.43.7
--
opensbi mailing list
opensbi@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/opensbi
next prev parent reply other threads:[~2026-09-01 2:22 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-01 2:22 [PATCH 0/3] Support OpenSBI firmware header Zong Li
2026-09-01 2:22 ` [PATCH 1/3] firmware: fw_base.S: add a formatted " Zong Li
2026-09-01 2:22 ` Zong Li [this message]
2026-09-01 2:22 ` [PATCH 3/3] docs: firmware: document the " Zong Li
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=20260901022211.2619560-3-zong.li@sifive.com \
--to=zong.li@sifive.com \
--cc=anup@brainfault.org \
--cc=opensbi@lists.infradead.org \
/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