Skip to content

Run a virtual iPhone on an Apple Silicon Mac.

Swift

Pin OpenAVB to upstream 3c7e7f6 and drop the carried patch

Jackyd9f2f1edbc54
diff --git a/AGENTS.md b/AGENTS.md
index 766c0b59..4bbedc3c 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -56,7 +56,7 @@ The `VPhone` scheme puts host programs in `VPhone.bundle/Contents/MacOS` and eve
### Key Patterns
-- New machines boot **OpenAVB**, not Apple's AVPBooter: `VPhone.bundle/Contents/Resources/OpenAVB/OpenAVB.bin`, checked in prebuilt under `VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/` from a pinned OpenAVB commit plus the patches in `patches/` beside it. Rebuild or check it only with `Build/UpdateOpenAVB.sh` (Homebrew LLVM/LLD, outside the repo). `fw patch` skips AVPBooter on an OpenAVB machine; `vphone-cli restore` uploads the iBSS identity first (`MobileRestoreCore/Recovery/openavb.c`); `vm config --rom` switches ROM. See `Research/Firmware/openavb_boot_rom.md`.
+- New machines boot **OpenAVB**, not Apple's AVPBooter: `VPhone.bundle/Contents/Resources/OpenAVB/OpenAVB.bin`, checked in prebuilt under `VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/` from a pinned OpenAVB commit plus any patches in `patches/` beside it (none at present). Rebuild or check it only with `Build/UpdateOpenAVB.sh` (Homebrew LLVM/LLD, outside the repo). `fw patch` skips AVPBooter on an OpenAVB machine; `vphone-cli restore` uploads the iBSS identity first (`MobileRestoreCore/Recovery/openavb.c`); `vm config --rom` switches ROM. See `Research/Firmware/openavb_boot_rom.md`.
- `vphone-cli` is the unentitled entry point. `VPhoneGuestLaunchPlanner` resolves `vphone-vm` beside the running executable, checks its entitlements, probes AMFI, and reports the exact allowlist command on refusal. It never obtains root itself.
- `vphone-escalator` manages AMFI cdhash admission only. It writes amfid heap state, not executable code. Root authorization belongs to the user or to `vphone-launchpad`; neither `vphone-cli` nor the bundle has an SMJobBless or sudo password flow.
diff --git a/Research/0_binary_patch_comparison.md b/Research/0_binary_patch_comparison.md
index b806844a..2bd5f621 100644
--- a/Research/0_binary_patch_comparison.md
+++ b/Research/0_binary_patch_comparison.md
@@ -3185,9 +3185,9 @@ they are.
## OpenAVB replaces AVPBooter on new machines (2026-10-18)
New machines boot OpenAVB (`Resources/OpenAVB/OpenAVB.bin`, built from
-`owngoal-dev/OpenAVB` at `115169e`, which includes vphone's five
-Virtualization.framework fixes, plus one carried patch that lets a guest
-reboot: the ROM waits for a SEP restarted by the reset) instead of
+`owngoal-dev/OpenAVB` at `3c7e7f6`, with no carried patches; upstream
+includes vphone's six Virtualization.framework fixes, the last of which lets a
+guest reboot: the ROM waits for a SEP restarted by the reset) instead of
Virtualization.framework's `AVPBooter.vresearch1.bin`. OpenAVB checks no image
digest, so `avpbooter-boot-dgst_bypass` has nothing to patch on such a machine:
`fw patch` drops the AVPBooter component when `romImages.avpBooter` names an
@@ -3205,6 +3205,7 @@ Two boot-chain files change form, not content:
No patch site moved and no patch was added or removed. The OpenAVB-side
changes (read-only ROM aperture, chip ID 0, DFU suffix CRC, remote-boot bit,
-USB device left running) were carried as patches at first and are now
-upstream (owngoal-dev/OpenAVB#1). `Firmware/openavb_boot_rom.md` describes
+USB device left running, SEP not ready after a warm reset) were carried as
+patches at first and are now upstream (owngoal-dev/OpenAVB#1 and #2).
+`Firmware/openavb_boot_rom.md` describes
each one and how it was found.
diff --git a/Research/Firmware/openavb_boot_rom.md b/Research/Firmware/openavb_boot_rom.md
index c0814428..be579fd8 100644
--- a/Research/Firmware/openavb_boot_rom.md
+++ b/Research/Firmware/openavb_boot_rom.md
@@ -78,8 +78,10 @@ Virtualization.framework it did not start. Five changes made it boot and
restore. vphone carried them as patches until upstream took them in
[owngoal-dev/OpenAVB#1](https://github.com/owngoal-dev/OpenAVB/pull/1),
merged as `02ca2e5`. Upstream then rewrote `main` into one commit, `115169e`,
-which keeps those changes, is the pin, and builds the same image the patched
-`54b67bf` did, and OpenAVB's `make verify` passes on it. Each change
+which keeps those changes and builds the same image the patched `54b67bf`
+did. A sixth change, found later, went upstream in
+[owngoal-dev/OpenAVB#2](https://github.com/owngoal-dev/OpenAVB/pull/2) as
+`3c7e7f6`, which is the pin; OpenAVB's `make verify` passes on it. Each change
was found on a live machine by the first thing that stopped, in this order:
1. **ROM aperture is read-only.** `_start` stores `avp_el2_present` at
@@ -109,9 +111,8 @@ was found on a live machine by the first thing that stopped, in this order:
(`4d682d940d6c950:514`: an event record whose type byte is not 1, 2 or 3).
The handoff now writes the device's two enable registers (`+0x404`,
`+0x400`) back to 0, and iBSS initializes it itself.
-6. **SEP not ready after a warm reset** (carried as
- `patches/0001-Wait-for-the-SEP-before-the-first-request-after-a-wa.patch`,
- not upstream yet). A guest reboot is a PSCI `SYSTEM_RESET`, which restarts
+6. **SEP not ready after a warm reset** (owngoal-dev/OpenAVB#2, carried as a
+ patch until it was merged). A guest reboot is a PSCI `SYSTEM_RESET`, which restarts
the SEP together with the AP. Cold boots and restores worked, but every
guest reboot hung: the console filled with `SEP RX timeout`, and
Launchpad's setup boot failed at "the guest did not come back from the
@@ -155,7 +156,7 @@ reach the VM's serial log.
accepted`, iBSS and iBEC print `Remote boot`, and the restore ends `Restore
Finished` with 86 boot-chain patches in the receipt.
With CFW installed it boots to iOS on OpenAVB and vphoned answers.
-- Guest reboot (2026-10-18, with patch 0001): `system.reboot` on that machine
+- Guest reboot (2026-10-18, with change 6): `system.reboot` on that machine
comes back in 10 s, three times out of three, with no `SEP RX timeout`.
`vm template setup` (Launchpad's setup boot, which failed at its reboot
step without the patch) passes, the reboot taking 9 s. A second full DFU
@@ -173,8 +174,14 @@ log show `OpenBootROM:` for OpenAVB.
## Upstream
-Changes 1 to 5 are upstream (owngoal-dev/OpenAVB#1; `main` at `115169e`
-carries them). Change 6 is carried in `Resources/OpenAVB/patches/` as a
-`git format-patch` file until upstream takes it; delete it when the pin moves
-past it.
+All six changes are upstream: 1 to 5 in owngoal-dev/OpenAVB#1 (folded into
+`115169e`), 6 in owngoal-dev/OpenAVB#2. vphone carries no patch. The pin is
+#2's own commit, `3c7e7f6`, not its merge `63c1ead`: both have the same tree
+and `main` reaches both, but upstream has rewritten `main` once, and GitHub
+keeps a PR's head commit reachable (`refs/pull/2/head`) after a rewrite. If
+upstream rewrites `main` again, `UpdateOpenAVB.sh` fails at checkout; move the
+pin to the rewritten commit that builds the same image.
+
+A change upstream has not taken yet goes in `Resources/OpenAVB/patches/` as a
+`git format-patch` file; delete it when the pin moves past it.
`UpdateOpenAVB.sh` refuses a patch that no longer applies.
diff --git a/VPhoneExecutable/VPhoneVirtualization/Build/UpdateOpenAVB.sh b/VPhoneExecutable/VPhoneVirtualization/Build/UpdateOpenAVB.sh
index 924a7965..3dd9dfdc 100755
--- a/VPhoneExecutable/VPhoneVirtualization/Build/UpdateOpenAVB.sh
+++ b/VPhoneExecutable/VPhoneVirtualization/Build/UpdateOpenAVB.sh
@@ -20,7 +20,7 @@
set -euo pipefail
repository="https://github.com/owngoal-dev/OpenAVB.git"
-pinned="115169e61a206a967bb2299e5c3145888741bba3"
+pinned="3c7e7f602d2d35af67f50454a8216337f35b11bb"
root="$(cd "${0:a:h}/../../.." && pwd)"
destination="$root/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB"
diff --git a/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/README.md b/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/README.md
index 45618260..9f68e0f8 100644
--- a/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/README.md
+++ b/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/README.md
@@ -10,8 +10,8 @@ the image into each new machine folder.
| | |
| --- | --- |
| Source | `https://github.com/owngoal-dev/OpenAVB` |
-| Commit | `115169e61a206a967bb2299e5c3145888741bba3` |
-| Patches | `patches/*.patch`, applied in order (one) |
+| Commit | `3c7e7f602d2d35af67f50454a8216337f35b11bb` |
+| Patches | none (`patches/*.patch`, applied in order, when there are any) |
| Target | `make build/OpenAVB.bin` |
| Toolchain | Homebrew LLVM 23.1.3 (`clang`, `llvm-objcopy`), Homebrew LLD 23.1.3 |
| SHA-256 | `d63ea62226d4383288a5a36c440aa2d1df8cccabe0118cac0ab2e0c23ba4c3d2` |
@@ -26,19 +26,18 @@ header). It clones and builds outside the repository and needs
## Patches
-One, not upstream yet:
-`0001-Wait-for-the-SEP-before-the-first-request-after-a-wa.patch`. After a
-guest reboot the SEP restarts together with the ROM, and a request sent
-before it is up was dropped, so every guest reboot hung. The ROM now waits
-for the SEP's first message before sending.
-
-OpenAVB's first commit, `54b67bf`, did not run under
+None. OpenAVB's first commit, `54b67bf`, did not run under
Virtualization.framework. Five changes found on live machines made it boot and
restore, and upstream took them in
-[owngoal-dev/OpenAVB#1](https://github.com/owngoal-dev/OpenAVB/pull/1). Upstream `main`
-(`115169e`) carries them and, without the patch above, builds the same image the patched
-`54b67bf` did. `Research/Firmware/openavb_boot_rom.md` has the evidence for
-each change.
+[owngoal-dev/OpenAVB#1](https://github.com/owngoal-dev/OpenAVB/pull/1), now
+part of `115169e`. A sixth let a guest reboot: after a reboot the SEP restarts
+together with the ROM, a request sent before it was up was dropped, and every
+reboot hung, so the ROM now waits for the SEP's first message before sending.
+Upstream took it in
+[owngoal-dev/OpenAVB#2](https://github.com/owngoal-dev/OpenAVB/pull/2). The pin
+is that PR's commit, `3c7e7f6`, which `main` reaches through the merge
+`63c1ead` (same tree), rather than the merge itself.
+`Research/Firmware/openavb_boot_rom.md` has the evidence for each change.
If vphone needs a change upstream has not taken yet, put it in `patches/` as a
`git format-patch` file. `UpdateOpenAVB.sh` applies patches in order and
diff --git a/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/patches/0001-Wait-for-the-SEP-before-the-first-request-after-a-wa.patch b/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/patches/0001-Wait-for-the-SEP-before-the-first-request-after-a-wa.patch
deleted file mode 100644
index 14b0fdb9..00000000
--- a/VPhoneExecutable/VPhoneVirtualization/Resources/OpenAVB/patches/0001-Wait-for-the-SEP-before-the-first-request-after-a-wa.patch
+++ /dev/null
@@ -1,88 +0,0 @@
-From 20ae8738b77e53cf49a3e1cba195cab449368274 Mon Sep 17 00:00:00 2001
-From: Jacky <[email protected]>
-Date: Sun, 18 Oct 2026 19:23:41 +0800
-Subject: [PATCH] Wait for the SEP before the first request after a warm reset
-
-On a guest reboot (PSCI SYSTEM_RESET) Virtualization.framework restarts
-the SEP together with the AP. OpenAVB got to the mailbox first, took it,
-and sent the bootstrap request at once. The mailbox dropped it
-("coprocessor.mailbox: Failed to send a message"), the SEP's own setup
-then cleared the AP's ownership, and each side waited for the other:
-the console filled with "SEP RX timeout" and the guest never came back.
-
-After a cold start the SEP is up long before the ROM gets there and its
-first message is already waiting. So after taking the slot, wait for
-that message, taking the slot back if the SEP's restart drops it, and
-send only then. Each ownership request also sets the enable bit at
-+0x10 first, as Apple's ROM and the SEP ROM do.
----
- src/protocol/sep/sep.c | 35 ++++++++++++++++++++++++++++++++++-
- 1 file changed, 34 insertions(+), 1 deletion(-)
-
-diff --git a/src/protocol/sep/sep.c b/src/protocol/sep/sep.c
-index 77c9cbc..6dba128 100644
---- a/src/protocol/sep/sep.c
-+++ b/src/protocol/sep/sep.c
-@@ -15,6 +15,7 @@ enum {
- SEP_TIMEOUT_RECOVERY_LIMIT = 5,
- SEP_TIMEOUT_POLL_INTERVAL_US = 100,
- SEP_FINALIZE_ROUNDS = 8,
-+ SEP_READY_TIMEOUT_US = 5000000,
- };
-
- static void early_console_write(const char *message) {
-@@ -29,11 +30,20 @@ static uint32_t timeout_recovery_count;
- static bool preflight_wait_complete;
- static uint8_t last_protocol_opcode;
-
--void sep_initialize(void) {
-+/*
-+ * Request the slot until it reads as owned, setting the enable bit at +0x10
-+ * before each request as Apple's ROM does, and as the SEP ROM does on its
-+ * side.
-+ */
-+static void sep_acquire(void) {
- uint32_t active_control = mmio_read32(SEP_MAILBOX_BASE + 0x14);
- while ((active_control & SEP_MAILBOX_OWNED) == 0) {
- /* The ownership register acknowledges writes, so an owned slot is
- * observed and preserved without issuing another ownership request. */
-+ mmio_write32(
-+ SEP_MAILBOX_BASE + 0x10,
-+ mmio_read32(SEP_MAILBOX_BASE + 0x10) | 1
-+ );
- mmio_write32(
- SEP_MAILBOX_BASE + 0x14,
- active_control | SEP_MAILBOX_OWNED
-@@ -47,6 +57,29 @@ void sep_initialize(void) {
- avp_memory_barrier();
- }
-
-+void sep_initialize(void) {
-+ sep_acquire();
-+ /*
-+ * Do not send before the SEP has announced itself. After a cold start it
-+ * is up long before this ROM gets here, and its first message is already
-+ * waiting. After a warm reset (a guest reboot) it restarts together with
-+ * this ROM and is slower: a request sent now is dropped by the mailbox,
-+ * and the SEP's own setup then clears the ownership taken above, so both
-+ * sides wait for each other. Wait for its message instead, taking the
-+ * slot back when its restart drops it.
-+ */
-+ const uint64_t start = avp_time_counter();
-+ const uint64_t limit = avp_time_microseconds_to_ticks(SEP_READY_TIMEOUT_US);
-+ while ((mmio_read32(SEP_MAILBOX_BASE + 0x1c) & 0x00020000) != 0) {
-+ if ((mmio_read32(SEP_MAILBOX_BASE + 0x14) & SEP_MAILBOX_OWNED) == 0) {
-+ sep_acquire();
-+ }
-+ if (avp_time_has_elapsed(start, limit)) {
-+ break;
-+ }
-+ }
-+}
-+
- static int32_t sep_receive(
- uint32_t timeout_microseconds,
- SEPMailboxMessage *message
---
-2.56.0
-