From e43257aa7d91ebc2bb46d0dcb2eb4ef51bbfc745 Mon Sep 17 00:00:00 2001
From: John Baldwin <jhb@FreeBSD.org>
Date: Thu, 1 Apr 2004 20:56:44 +0000
Subject: [PATCH] Finish fixing up Alpha to work with an MP safe ptrace(): -
 ptrace_single_step() is no longer called with the proc lock held, so   don't
 try to unlock it and then relock it. - Push Giant down into proc_rwmem()
 instead of forcing all the consumers   (including Alpha breakpoint support)
 to explicitly wrap calls to   proc_rwmem() with Giant.

Tested by:	kensmith
---
 sys/alpha/alpha/machdep.c |  9 +++------
 sys/kern/sys_process.c    | 16 ++++++++--------
 2 files changed, 11 insertions(+), 14 deletions(-)

diff --git a/sys/alpha/alpha/machdep.c b/sys/alpha/alpha/machdep.c
index cff99c398e5..b374cbfb0d2 100644
--- a/sys/alpha/alpha/machdep.c
+++ b/sys/alpha/alpha/machdep.c
@@ -1914,10 +1914,9 @@ ptrace_single_step(struct thread *td)
 	if (td->td_md.md_flags & (MDTD_STEP1|MDTD_STEP2))
 		panic("ptrace_single_step: step breakpoints not removed");
 
-	PROC_UNLOCK(td->td_proc);
 	error = ptrace_read_int(td, pc, &ins.bits);
 	if (error)
-		goto err;
+		return (error);
 
 	switch (ins.branch_format.opcode) {
 
@@ -1957,20 +1956,18 @@ ptrace_single_step(struct thread *td)
 	td->td_md.md_sstep[0].addr = addr[0];
 	error = ptrace_set_bpt(td, &td->td_md.md_sstep[0]);
 	if (error)
-		goto err;
+		return (error);
 	if (count == 2) {
 		td->td_md.md_sstep[1].addr = addr[1];
 		error = ptrace_set_bpt(td, &td->td_md.md_sstep[1]);
 		if (error) {
 			ptrace_clear_bpt(td, &td->td_md.md_sstep[0]);
-			goto err;
+			return (error);
 		}
 		td->td_md.md_flags |= MDTD_STEP2;
 	} else
 		td->td_md.md_flags |= MDTD_STEP1;
 
-err:
-	PROC_LOCK(td->td_proc);
 	return (error);
 }
 
diff --git a/sys/kern/sys_process.c b/sys/kern/sys_process.c
index dcfad3d0cca..a43e7b0a089 100644
--- a/sys/kern/sys_process.c
+++ b/sys/kern/sys_process.c
@@ -154,18 +154,21 @@ proc_rwmem(struct proc *p, struct uio *uio)
 	vm_prot_t reqprot;
 	int error, writing;
 
-	GIANT_REQUIRED;
-
+	mtx_lock(&Giant);
 	/*
 	 * if the vmspace is in the midst of being deallocated or the
 	 * process is exiting, don't try to grab anything.  The page table
 	 * usage in that process can be messed up.
 	 */
 	vm = p->p_vmspace;
-	if ((p->p_flag & P_WEXIT))
+	if ((p->p_flag & P_WEXIT)) {
+		mtx_unlock(&Giant);
 		return (EFAULT);
-	if (vm->vm_refcnt < 1)
+	}
+	if (vm->vm_refcnt < 1) {
+		mtx_unlock(&Giant);
 		return (EFAULT);
+	}
 	++vm->vm_refcnt;
 	/*
 	 * The map we want...
@@ -274,6 +277,7 @@ proc_rwmem(struct proc *p, struct uio *uio)
 	} while (error == 0 && uio->uio_resid > 0);
 
 	vmspace_free(vm);
+	mtx_unlock(&Giant);
 	return (error);
 }
 
@@ -602,9 +606,7 @@ kern_ptrace(struct thread *td, int req, pid_t pid, void *addr, int data)
 		uio.uio_segflg = UIO_SYSSPACE;	/* i.e.: the uap */
 		uio.uio_rw = write ? UIO_WRITE : UIO_READ;
 		uio.uio_td = td;
-		mtx_lock(&Giant);
 		error = proc_rwmem(p, &uio);
-		mtx_unlock(&Giant);
 		if (uio.uio_resid != 0) {
 			/*
 			 * XXX proc_rwmem() doesn't currently return ENOSPC,
@@ -645,9 +647,7 @@ kern_ptrace(struct thread *td, int req, pid_t pid, void *addr, int data)
 		default:
 			return (EINVAL);
 		}
-		mtx_lock(&Giant);
 		error = proc_rwmem(p, &uio);
-		mtx_unlock(&Giant);
 		piod->piod_len -= uio.uio_resid;
 		return (error);