netfs: zero the tail of a short DIO/unbuffered readgrafted

The buffered read collector zero-fills the tail of a short read that
stops below the inode's i_size (netfs_clear_unread()), so a read that
races an extending write still returns the expected number of bytes.
The non-buffered collector path does no such thing: it just records
how much was transferred.

Add the same zero-fill for the non-buffered case, gated on
NETFS_SREQ_CLEAR_TAIL: a subreq's source sets that flag when a short
result from it is known to be safe to treat as a hole, as opposed to
NETFS_SREQ_HIT_EOF, which means the read genuinely ran off the end of
the file and should be reported short as-is. Only CLEAR_TAIL should
zero-fill here; a real EOF must stay a real short read.

No source currently sets CLEAR_TAIL on an unbuffered/DIO subrequest,
so this is inert on its own -- a following change teaches cifs to set
it in the one case that needs it.

Fixes: e2d46f2ec332 ("netfs: Change the read result collector to only use one work item")
Reviewed-by: David Howells <dhowells@redhat.com>
Reviewed-by: Namjae Jeon <linkinjeon@kernel.org>
Signed-off-by: Paulo Alcantara <pc@manguebit.org>
Cc: Christian Brauner <brauner@kernel.org>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Ronnie Sahlberg <ronniesahlberg@gmail.com>
Cc: Shyam Prasad N <sprasad@microsoft.com>
Cc: Tom Talpey <tom@talpey.com>
Cc: Bharath SM <bharathsm@microsoft.com>
Cc: stable@vger.kernel.org

1 files changed, 88 insertions, 0 deletions

diff --git a/lib/syscall.c b/lib/syscall.c
new file mode 100644
index 000000000..006e256d2
--- /dev/null
+++ b/lib/syscall.c
@@ -0,0 +1,88 @@
+// SPDX-License-Identifier: GPL-2.0
+#include <linux/ptrace.h>
+#include <linux/sched.h>
+#include <linux/sched/task_stack.h>
+#include <linux/export.h>
+#include <asm/syscall.h>
+
+static int collect_syscall(struct task_struct *target, struct syscall_info *info)
+{
+	unsigned long args[6] = { };
+	struct pt_regs *regs;
+
+	if (!try_get_task_stack(target)) {
+		/* Task has no stack, so the task isn't in a syscall. */
+		memset(info, 0, sizeof(*info));
+		info->data.nr = -1;
+		return 0;
+	}
+
+	regs = task_pt_regs(target);
+	if (unlikely(!regs)) {
+		put_task_stack(target);
+		return -EAGAIN;
+	}
+
+	info->sp = user_stack_pointer(regs);
+	info->data.instruction_pointer = instruction_pointer(regs);
+
+	info->data.nr = syscall_get_nr(target, regs);
+	if (info->data.nr != -1L)
+		syscall_get_arguments(target, regs, args);
+
+	info->data.args[0] = args[0];
+	info->data.args[1] = args[1];
+	info->data.args[2] = args[2];
+	info->data.args[3] = args[3];
+	info->data.args[4] = args[4];
+	info->data.args[5] = args[5];
+
+	put_task_stack(target);
+	return 0;
+}
+
+/**
+ * task_current_syscall - Discover what a blocked task is doing.
+ * @target:		thread to examine
+ * @info:		structure with the following fields:
+ *			 .sp        - filled with user stack pointer
+ *			 .data.nr   - filled with system call number or -1
+ *			 .data.args - filled with @maxargs system call arguments
+ *			 .data.instruction_pointer - filled with user PC
+ *
+ * If @target is blocked in a system call, returns zero with @info.data.nr
+ * set to the call's number and @info.data.args filled in with its
+ * arguments. Registers not used for system call arguments may not be available
+ * and it is not kosher to use &struct user_regset calls while the system
+ * call is still in progress.  Note we may get this result if @target
+ * has finished its system call but not yet returned to user mode, such
+ * as when it's stopped for signal handling or syscall exit tracing.
+ *
+ * If @target is blocked in the kernel during a fault or exception,
+ * returns zero with *@info.data.nr set to -1 and does not fill in
+ * @info.data.args. If so, it's now safe to examine @target using
+ * &struct user_regset get() calls as long as we're sure @target won't return
+ * to user mode.
+ *
+ * Returns -%EAGAIN if @target does not remain blocked.
+ */
+int task_current_syscall(struct task_struct *target, struct syscall_info *info)
+{
+	unsigned long ncsw;
+	unsigned int state;
+
+	if (target == current)
+		return collect_syscall(target, info);
+
+	state = READ_ONCE(target->__state);
+	if (unlikely(!state))
+		return -EAGAIN;
+
+	ncsw = wait_task_inactive(target, state);
+	if (unlikely(!ncsw) ||
+	    unlikely(collect_syscall(target, info)) ||
+	    unlikely(wait_task_inactive(target, state) != ncsw))
+		return -EAGAIN;
+
+	return 0;
+}