v1.34-notify-1: rebase onto scoutfs v1.34

Upstream v1.34 added iomap-based buffered and direct I/O, splitting
kmod/src/file.c into a KC_USE_IOMAP_FOR_IO arm and an #else arm, each
with its own scoutfs_file_read_iter.  Patch 0002 now hooks both.  EL8
and EL9 kernels lack copy_page_to_iter_nofault (v6.4+) and compile the
#else arm, so the shipped kmods are unaffected either way, but hooking
only that arm would make READ notifications a silent no-op on any
newer kernel that later enters the build matrix.

Patch 0001 and 0003 are unchanged apart from context offsets; the
ioctl ABI is untouched (SCOUTFS_IOC_READ_NOTIFY stays nr 26 -- v1.34
did not renumber).

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

README.md

@@ -10,24 +10,32 @@ currently-supported upstream scoutfs version.
 Three patches against each supported scoutfs release:
 
 1. **Kmod core** — a per-mount 64 KiB ring of 64-byte notification records and
-   a single-reader drain ioctl (`SCOUTFS_IOC_READ_NOTIFY`, nr 25).  Emit is
+   a single-reader drain ioctl (`SCOUTFS_IOC_READ_NOTIFY`, nr 26).  Emit is
    non-blocking, drop-on-full; the monotonic `seq` field exposes drops to
    consumers.  Three percpu counters (`notify_emitted`,
    `notify_dropped_ring_full`, `notify_reader_attached`).  No mount option,
    no sysfs toggle.
 2. **Kmod hooks** — `->open` wrapper and READ emit in
-   `scoutfs_file_aio_read` / `scoutfs_file_read_iter`.  Every hook is a
-   single predicted-false branch when no reader is attached.  Nothing in
-   the data-waiter state machine is touched.
+   `scoutfs_file_read_iter`.  Every hook is a single predicted-false
+   branch when no reader is attached.  Nothing in the data-waiter state
+   machine is touched.  (Branches up to `v1.32` also hooked
+   `scoutfs_file_aio_read`; scoutfs v1.33 removed the
+   `KC_LINUX_HAVE_FOP_AIO_READ` compat paths, so `read_iter` is now the
+   only read path.  scoutfs v1.34 added iomap buffered/direct I/O, which
+   splits `scoutfs_file_read_iter` into a `KC_USE_IOMAP_FOR_IO` arm and
+   an `#else` arm; from `v1.34` on both arms carry the hook.)
 3. **scoutfs-notifyd (Go)** — userspace daemon that binds
    `/run/scoutfs/<fsid>/notify.sock` (AF_UNIX SOCK_SEQPACKET, mode 0600,
    root-only), drains the ring, and broadcasts each record to connected
    clients.  Pure-stdlib Go; no external module dependencies.  Shipped
    with a systemd template unit `scoutfs-notifyd@<mountpoint>.service`.
 
-The notify ABI (`SCOUTFS_IOC_READ_NOTIFY` nr 25, 64-byte event record) is
-identical across all three supported scoutfs versions, so consumers and
-the daemon do not need per-version branching in their own code.
+The 64-byte event record layout is identical on every branch, so
+consumers and the daemon do not need per-version branching in their own
+code.  The ioctl number is not: it is nr 25 on `v1.28`-`v1.30` and nr 26
+from `v1.32` on, because upstream v1.32 took nr 25 for
+`SCOUTFS_IOC_INJECT_TOTL_DELTA`.  Neither scoutfs v1.33 nor v1.34
+renumbered again.
 
 ## Repository layout
 
@@ -39,7 +47,10 @@ NOT apply cross-version without rebasing.
 |----------------|-----------------|------------------|
 | `v1.28`        | `v1.28`         | `v1.28-notify-1` |
 | `v1.29`        | `v1.29`         | `v1.29-notify-1` |
-| `main` / `v1.30` | `v1.30`       | `v1.30-notify-1` |
+| `v1.30`        | `v1.30`         | `v1.30-notify-1` |
+| `v1.32`        | `v1.32`         | `v1.32-notify-1` |
+| `v1.33`        | `v1.33`         | `v1.33-notify-1` |
+| `v1.34`        | `v1.34`         | `v1.34-notify-1` |
 
 `main` tracks the newest supported scoutfs version.  When a new scoutfs
 release ships, a new branch is created and the patches are rebased onto

base.txt

@@ -1 +1 @@
-v1.32
+v1.34

patches/0001-notify-core-file-access-notification-infrastructure.patch

@@ -1,4 +1,4 @@
-From 50eba28d24c3b98c71d851c8093abc27a1835cce Mon Sep 17 00:00:00 2001
+From e01e070f486cf0f525cf4300464201e092b62e4d Mon Sep 17 00:00:00 2001
 From: William Gill <claude@williamgill.net>
 Date: Wed, 22 Apr 2026 14:38:51 -0500
 Subject: [PATCH 1/3] notify: core file-access notification infrastructure
@@ -58,10 +58,10 @@ follow-up patch.
  create mode 100644 kmod/src/notify.h
 
 diff --git a/kmod/src/Makefile b/kmod/src/Makefile
-index fa632aa..de14c74 100644
+index 7fd7319..c54b47e 100644
 --- a/kmod/src/Makefile
 +++ b/kmod/src/Makefile
-@@ -31,6 +31,7 @@ scoutfs-y +=			\
+@@ -32,6 +32,7 @@ scoutfs-y +=			\
  	lock_server.o		\
  	msg.o			\
  	net.o			\
@@ -70,10 +70,10 @@ index fa632aa..de14c74 100644
  	options.o		\
  	per_task.o		\
 diff --git a/kmod/src/counters.h b/kmod/src/counters.h
-index 9088496..94e4f44 100644
+index cb7a2e1..83c0976 100644
 --- a/kmod/src/counters.h
 +++ b/kmod/src/counters.h
-@@ -156,6 +156,9 @@
+@@ -157,6 +157,9 @@
  	EXPAND_COUNTER(net_recv_invalid_message)		\
  	EXPAND_COUNTER(net_recv_messages)			\
  	EXPAND_COUNTER(net_unknown_request)			\
@@ -84,7 +84,7 @@ index 9088496..94e4f44 100644
  	EXPAND_COUNTER(orphan_scan_attempts)			\
  	EXPAND_COUNTER(orphan_scan_cached)			\
 diff --git a/kmod/src/ioctl.c b/kmod/src/ioctl.c
-index 156538a..4740ff0 100644
+index 6da71d9..5703c6a 100644
 --- a/kmod/src/ioctl.c
 +++ b/kmod/src/ioctl.c
 @@ -47,6 +47,7 @@
@@ -94,7 +94,7 @@ index 156538a..4740ff0 100644
 +#include "notify.h"
  #include "scoutfs_trace.h"
  #include "util.h"
- 
+ #include "msg.h"
 @@ -1829,6 +1830,8 @@ long scoutfs_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
  		return scoutfs_ioc_punch_offline(file, arg);
  	case SCOUTFS_IOC_INJECT_TOTL_DELTA:
@@ -622,7 +622,7 @@ index 0000000..fe6a2d9
 +
 +#endif /* _SCOUTFS_NOTIFY_H_ */
 diff --git a/kmod/src/super.c b/kmod/src/super.c
-index 3c83716..3c028b0 100644
+index f2e1420..94922be 100644
 --- a/kmod/src/super.c
 +++ b/kmod/src/super.c
 @@ -51,6 +51,7 @@

patches/0002-notify-file-open-read-hook-sites.patch

@@ -1,10 +1,7 @@
-From 482c0be372e2982cd7ca9ad7c0ea3522437886c6 Mon Sep 17 00:00:00 2001
+From cf9d03234f1424b95d08e3a3492479f550aee455 Mon Sep 17 00:00:00 2001
 From: William Gill <claude@williamgill.net>
 Date: Wed, 22 Apr 2026 14:40:04 -0500
 Subject: [PATCH 2/3] notify: file open/read hook sites
-MIME-Version: 1.0
-Content-Type: text/plain; charset=UTF-8
-Content-Transfer-Encoding: 8bit
 
 Wires the notification emit path into the two scoutfs file
 operations that carry user-visible activity we want to observe.
@@ -13,18 +10,25 @@ OPEN (data.c):
   A new scoutfs_file_open() wrapper is installed as ->open in
   scoutfs_file_fops.  The wrapper calls generic_file_open() to
   preserve existing VFS default semantics for regular files,
-  then — only on success and only when notifications are enabled
-  — emits a SCOUTFS_NOTIFY_TYPE_OPEN record.  File mode is
+  then - only on success and only when notifications are enabled
+  - emits a SCOUTFS_NOTIFY_TYPE_OPEN record.  File mode is
   inspected for FMODE_WRITE to set SCOUTFS_NOTIFY_F_WRITE_OPEN.
 
 READ (file.c):
-  Both the aio_read (KC_LINUX_HAVE_FOP_AIO_READ) and read_iter
-  paths get an emit placed past the existing data-waiter retry
-  check, guarded on (ret > 0) so only successful reads are
-  reported and retries never double-count.  start_pos is captured
-  at entry of read_iter before generic_file_read_iter advances
+  Each scoutfs_file_read_iter() gets an emit placed past the
+  existing data-waiter retry check, guarded on (ret > 0) so only
+  successful reads are reported and retries never double-count.
+  start_pos is captured at function entry before the read advances
   iocb->ki_pos.
 
+Rebased onto v1.34: upstream added iomap-based buffered and direct
+I/O, so file.c now carries two scoutfs_file_read_iter definitions -
+one under KC_USE_IOMAP_FOR_IO and the pre-v1.34 one in the #else
+arm.  Both get the same hook in the same position.  EL8 and EL9
+kernels lack copy_page_to_iter_nofault (v6.4+) and so compile the
+#else arm, but hooking only that arm would leave notifications a
+silent no-op on any newer kernel that later enters the build matrix.
+
 Every hook is behind unlikely(READ_ONCE(sbi->notify_enabled)), so
 when no userspace reader is attached the hook reduces to a single
 predicted-false branch.  No scoutfs state is mutated, no error is
@@ -33,22 +37,22 @@ propagated, and no existing control flow is altered.
 Nothing in the data-waiter state machine is touched.
 ---
  kmod/src/data.c | 28 ++++++++++++++++++++++++++++
- kmod/src/file.c | 11 +++++++++++
- 2 files changed, 39 insertions(+)
+ kmod/src/file.c | 12 ++++++++++++
+ 2 files changed, 40 insertions(+)
 
 diff --git a/kmod/src/data.c b/kmod/src/data.c
-index b6ca5eb..439ce09 100644
+index 44ca4c9..3fea851 100644
 --- a/kmod/src/data.c
 +++ b/kmod/src/data.c
-@@ -41,6 +41,7 @@
- #include "msg.h"
+@@ -45,6 +45,7 @@
  #include "ext.h"
  #include "util.h"
+ #include "iomap.h"
 +#include "notify.h"
  
  /*
   * We want to amortize work done after dirtying the shared transaction
-@@ -2312,6 +2313,32 @@ const struct address_space_operations scoutfs_file_aops = {
+@@ -2254,11 +2255,38 @@ const struct address_space_operations scoutfs_file_aops = {
  	.write_end		= scoutfs_write_end,
  };
  
@@ -79,40 +83,46 @@ index b6ca5eb..439ce09 100644
 +}
 +
  const struct file_operations scoutfs_file_fops = {
- #ifdef KC_LINUX_HAVE_FOP_AIO_READ
- 	.read		= do_sync_read,
-@@ -2324,6 +2351,7 @@ const struct file_operations scoutfs_file_fops = {
+ 	.read_iter	= scoutfs_file_read_iter,
+ 	.write_iter	= scoutfs_file_write_iter,
  	.splice_read	= generic_file_splice_read,
  	.splice_write	= iter_file_splice_write,
- #endif
 +	.open		= scoutfs_file_open,
  	.mmap		= scoutfs_file_mmap,
  	.unlocked_ioctl	= scoutfs_ioctl,
  	.fsync		= scoutfs_file_fsync,
 diff --git a/kmod/src/file.c b/kmod/src/file.c
-index 15158a2..44f1bc4 100644
+index 730a662..cf54871 100644
 --- a/kmod/src/file.c
 +++ b/kmod/src/file.c
-@@ -29,6 +29,7 @@
- #include "per_task.h"
- #include "omap.h"
- #include "quota.h"
+@@ -33,6 +33,7 @@
+ #include "iomap.h"
+ #include "trans.h"
+ #include "msg.h"
 +#include "notify.h"
  
- #ifdef KC_LINUX_HAVE_FOP_AIO_READ
- /*
-@@ -84,6 +85,10 @@ out:
+ #ifdef KC_USE_IOMAP_FOR_IO
+ 
+@@ -187,6 +188,7 @@ ssize_t scoutfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
+ 	struct scoutfs_lock *scoutfs_inode_lock;
+ 	SCOUTFS_DECLARE_PER_TASK_ENTRY(pt_data_ent);
+ 	DECLARE_DATA_WAIT(dw);
++	loff_t start_pos = iocb->ki_pos;
+ 	int lock_flags = SCOUTFS_LKF_REFRESH_INODE;
+ 	bool is_dio = (iocb->ki_flags & IOCB_DIRECT);
+ 	bool nowait = (iocb->ki_flags & IOCB_NOWAIT);
+@@ -258,6 +260,10 @@ out:
  			goto retry;
  	}
  
 +	if (ret > 0 && unlikely(READ_ONCE(SCOUTFS_SB(sb)->notify_enabled)))
 +		scoutfs_notify_emit(sb, SCOUTFS_NOTIFY_TYPE_READ,
-+				    scoutfs_ino(inode), pos, ret, 0);
++				    scoutfs_ino(inode), start_pos, ret, 0);
 +
  	return ret;
  }
  
-@@ -166,6 +171,7 @@ ssize_t scoutfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
+@@ -578,6 +584,7 @@ ssize_t scoutfs_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
  	struct scoutfs_lock *scoutfs_inode_lock = NULL;
  	SCOUTFS_DECLARE_PER_TASK_ENTRY(pt_ent);
  	DECLARE_DATA_WAIT(dw);
@@ -120,7 +130,7 @@ index 15158a2..44f1bc4 100644
  	int ret;
  
  retry:
-@@ -200,6 +206,11 @@ out:
+@@ -612,6 +619,11 @@ out:
  		if (ret == 0)
  			goto retry;
  	}

patches/0003-notify-scoutfs-notifyd-userspace-relay-daemon-Go-1.2.patch

@@ -1,4 +1,4 @@
-From 31ef1c5ff1522a6de4501ff41fed0d72ecb57e8d Mon Sep 17 00:00:00 2001
+From f95104cb0b1a46cfc2dbd7a9573eb80ff9176445 Mon Sep 17 00:00:00 2001
 From: William Gill <claude@williamgill.net>
 Date: Wed, 22 Apr 2026 14:40:34 -0500
 Subject: [PATCH 3/3] notify: scoutfs-notifyd userspace relay daemon (Go 1.26)