Am 04.02.2007 02:56 schrieb Andrew Morton:

On Sun, 04 Feb 2007 02:32:41 +0100 Tilman Schmidt [off-list ref] wrote:

quoted

quoted

quoted

+	spin_lock_irqsave(&cs->cmdlock, flags);
+	cb = cs->cmdbuf;
+	spin_unlock_irqrestore(&cs->cmdlock, flags);

It is doubtful if the locking here does anything useful.

It assures atomicity when reading the cs->cmdbuf pointer.

I think it's bogus.  If the quantity being copied here is more than 32-bits
then yes, a lock is appropriate.  But if it's a single word then it's
unlikely that the locking does anything useful.  Or there might be a bug
here.

It's a pointer. Are reads and writes of pointer sized objects
guaranteed to be atomic on every platform? If so, I'll happily
omit the locking.

quoted

quoted

quoted

+	spin_lock_irqsave(&cs->cmdlock, flags);
+	cb->prev = cs->lastcmdbuf;
+	if (cs->lastcmdbuf)
+		cs->lastcmdbuf->next = cb;
+	else {
+		cs->cmdbuf = cb;
+		cs->curlen = len;
+	}
+	cs->cmdbytes += len;
+	cs->lastcmdbuf = cb;
+	spin_unlock_irqrestore(&cs->cmdlock, flags);

Would the use of list_heads simplify things here?

I don't think so. The operations in list.h do not keep track of
the total byte count, and adding that in a race-free way appears
non-trivial.

Maintaining a byte count isn't related to maintaining a list.

Sure. But your question was whether the list.h operations would
simplify this code. AFAICS it wouldn't, because the necessity
of maintaining the byte count would complicate a list.h based
solution beyond the current one. Also, this is part of the
interface with the components of the Gigaset driver which are
already part of the kernel. Changing this to a list_head now
would require significant changes in those other parts, too.

quoted

quoted

quoted

+	tail = atomic_read(&inbuf->tail);
+	head = atomic_read(&inbuf->head);
+	gig_dbg(DEBUG_INTR, "buffer state: %u -> %u, receive %u bytes",
+		head, tail, count);
+
+	if (head <= tail) {
+		n = RBUFSIZE - tail;
+		if (count >= n) {
+			/* buffer wraparound */
+			memcpy(inbuf->data + tail, buf, n);
+			tail = 0;
+			buf += n;
+			count -= n;
+		} else {
+			memcpy(inbuf->data + tail, buf, count);
+			tail += count;
+			buf += count;
+			count = 0;
+		}
+	}

Perhaps the (fairly revolting) circ_buf.h can be used for this stuff.

It probably could, but IMHO readability would suffer rather than improve.

How about kernel/kfifo.c?

That would indeed fit the bill. But again, this code matches
parts of drivers/isdn/gigaset which are already in the kernel,
and changing it here would require significant corresponding
changes in those other parts.

I'll gladly consider your last two propositions (list_head for
cs->lastcmdbuf and kfifo for cs->inbuf) for a future revision of
the entire set of drivers in drivers/isdn/gigaset, but it goes
way beyond the scope of the present patch, which merely aims at
adding the missing M101 hardware driver.

Thanks,
Tilman

-- 
Tilman Schmidt                          E-Mail: tilman@imap.cc
Bonn, Germany
Diese Nachricht besteht zu 100% aus wiederverwerteten Bits.
Ungeöffnet mindestens haltbar bis: (siehe Rückseite)