[syscalls][docs] Fix a few DESCRIPTIONs that don't match SYNOPSIS

https://fuchsia-review.googlesource.com/c/zircon/+/233524/ wishes that
our documents were more structured than they really are in practice, so
it's not currently landable.

But manually looking through the warnings it emits points out a few
mistakes, fixed here.

ZX-2399 #comment [syscalls][docs] Fix a few DESCRIPTIONs that don't match SYNOPSIS
ZX-3106 #comment [syscalls][docs] Fix a few DESCRIPTIONs that don't match SYNOPSIS

Test: CQ
Change-Id: I0ce85bec4d5d2e33adf1d06dee3a7ac883897c21
diff --git a/docs/syscalls/fifo_write.md b/docs/syscalls/fifo_write.md
index 4e82573..c91b542 100644
--- a/docs/syscalls/fifo_write.md
+++ b/docs/syscalls/fifo_write.md
@@ -23,7 +23,7 @@
 ## DESCRIPTION
 
 **fifo_write**() attempts to write up to *count* elements
-(*count * elem_size* bytes) from *buf* to the fifo specified by *handle*.
+(*count * elem_size* bytes) from *data* to the fifo specified by *handle*.
 
 Fewer elements may be written than requested if there is insufficient
 room in the fifo to contain all of them. The number of
diff --git a/docs/syscalls/handle_close_many.md b/docs/syscalls/handle_close_many.md
index 61ca1d4..441c4b4 100644
--- a/docs/syscalls/handle_close_many.md
+++ b/docs/syscalls/handle_close_many.md
@@ -19,11 +19,11 @@
 
 ## DESCRIPTION
 
-**handle_close_many**() closes a number of *handle*s, causing each
+**handle_close_many**() closes a number of handles, causing each
 underlying object to be reclaimed by the kernel if no other handles to
 it exist.
 
-If a *handle* was used in a pending [object_wait_one](object_wait_one.md) or a
+If a handle was used in a pending [object_wait_one](object_wait_one.md) or a
 [object_wait_many](object_wait_many.md) call, the wait will be aborted.
 
 This operation closes all handles presented to it, even if one or more
diff --git a/docs/syscalls/object_get_property.md b/docs/syscalls/object_get_property.md
index f1b1aaf..55a0664 100644
--- a/docs/syscalls/object_get_property.md
+++ b/docs/syscalls/object_get_property.md
@@ -31,8 +31,8 @@
 have the prefix **ZX_PROP_**, and are described below.
 
 The *value* parameter holds the property value, and must be a pointer to a
-buffer of *size* bytes. Different properties expect different value types/sizes
-as described below.
+buffer of *value_size* bytes. Different properties expect different value
+types/sizes as described below.
 
 ## PROPERTIES
 
diff --git a/docs/syscalls/process_read_memory.md b/docs/syscalls/process_read_memory.md
index 0d4ff99..1b755d6 100644
--- a/docs/syscalls/process_read_memory.md
+++ b/docs/syscalls/process_read_memory.md
@@ -34,7 +34,7 @@
 enough for at least this many bytes. *buffer_size* must be greater than zero
 and less than or equal to 64MB.
 
-*actual_size* the actual number of bytes read is stored here. Less bytes than
+*actual* the actual number of bytes read is stored here. Less bytes than
 requested may be returned if *vaddr*+*buffer_size* extends beyond the memory
 mapped in the process.
 
diff --git a/docs/syscalls/vcpu_create.md b/docs/syscalls/vcpu_create.md
index f38e77f..e0c6e9d 100644
--- a/docs/syscalls/vcpu_create.md
+++ b/docs/syscalls/vcpu_create.md
@@ -29,7 +29,7 @@
 *entry* is the instruction pointer used to indicate where in guest physical
 memory execution of the VCPU should start.
 
-*vcpu* is bound to the thread that created it, and all syscalls that operate on
+*out* is bound to the thread that created it, and all syscalls that operate on
 it must be called from the same thread, with the exception of
 **vcpu_interrupt**().
 
diff --git a/docs/syscalls/vmar_protect.md b/docs/syscalls/vmar_protect.md
index 9deadb4..54eae3c3 100644
--- a/docs/syscalls/vmar_protect.md
+++ b/docs/syscalls/vmar_protect.md
@@ -24,16 +24,16 @@
 **vmar_protect**() alters the access protections for the memory mappings
 in the range of *len* bytes starting from *addr*. The *options* argument should
 be a bitwise-or of one or more of the following:
-- **ZX_VM_PERM_READ**  Map as readable.  It is an error if *vmar*
-  does not have *ZX_VM_CAN_MAP_READ* permissions or the *vmar* handle does
+- **ZX_VM_PERM_READ**  Map as readable.  It is an error if *handle*
+  does not have *ZX_VM_CAN_MAP_READ* permissions or *handle* does
   not have the *ZX_RIGHT_READ* right.  It is also an error if the VMO handle
   used to create the mapping did not have the *ZX_RIGHT_READ* right.
-- **ZX_VM_PERM_WRITE**  Map as writable.  It is an error if *vmar*
-  does not have *ZX_VM_CAN_MAP_WRITE* permissions or the *vmar* handle does
+- **ZX_VM_PERM_WRITE**  Map as writable.  It is an error if *handle*
+  does not have *ZX_VM_CAN_MAP_WRITE* permissions or *handle* does
   not have the *ZX_RIGHT_WRITE* right.  It is also an error if the VMO handle
   used to create the mapping did not have the *ZX_RIGHT_WRITE* right.
-- **ZX_VM_PERM_EXECUTE**  Map as executable.  It is an error if *vmar*
-  does not have *ZX_VM_CAN_MAP_EXECUTE* permissions or the *vmar* handle does
+- **ZX_VM_PERM_EXECUTE**  Map as executable.  It is an error if *handle*
+  does not have *ZX_VM_CAN_MAP_EXECUTE* permissions or *handle* does
   not have the *ZX_RIGHT_EXECUTE* right.  It is also an error if the VMO handle
   used to create the mapping did not have the *ZX_RIGHT_EXECUTE* right.