Thread (9 messages) 9 messages, 3 authors, 2006-12-20

Re: [PATCH 6/15] hypervisor console driver for Celleb

From: Ishizaki Kou <hidden>
Date: 2006-12-14 01:42:49

Linas-san,

Thanks for your comment.
On Tue, Dec 12, 2006 at 12:31:29PM +0900, Ishizaki Kou wrote:
quoted
+
+static int hvc_beat_get_chars(uint32_t vtermno, char *buf, int cnt)
+{
+	unsigned long kb[2];
+	unsigned long got;
+
+	if (beat_get_term_char(vtermno, &got, &kb[0], &kb[1]) == 0) {
+	   memcpy(buf, kb, got);
+		       return got;
This seems to completely ignore "cnt". Thus, I presume that
beat_get_term_char might return more chars than there is room for in buf,
thus corrupting something, somewhere.
This depends "beat_get_term_char" returns only one character at once
(for now), and assumes cnt > 0. This assumption will reduce code for
now.
quoted
+static int hvc_beat_put_chars(uint32_t vtermno, const char *buf, int cnt)
+{
+	unsigned long kb[2];
+
+	memcpy(kb, buf, sizeof(kb));
+	beat_put_term_char(vtermno, cnt, kb[0], kb[1]);
+	return cnt;
+}
I can't imagine how this can possibly work. 
What if "cnt" is greater than 8?
This routine assumes that 0 <= cnt <= 16, that is already checked by
caller. (Note that "unsigned long" is 8 bytes long at ppc64)

Best regards,
Kou Ishizaki
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help