summaryrefslogtreecommitdiff
path: root/dix
diff options
context:
space:
mode:
authorJamey Sharp <jamey@minilop.net>2010-05-06 21:39:43 -0700
committerJamey Sharp <jamey@minilop.net>2010-05-12 18:09:55 -0700
commit65cedf39054cf3a1e695e84ac228cce9e8d48097 (patch)
tree9a0b41cf70562a6e9e06263486f2fcf1552da46a /dix
parentafcbbd6dfea51c5b0adca0d720edc02ba0c2dc16 (diff)
Kill DoChangeGC in favor of dixChangeGC.
This doesn't change any behavior, but it isn't clear whether NullClient is correct in all cases. As ajax says, > For most of these changes, I think it's correct to use NullClient, > since they are server-initiated changes and should not fail for (eg) > xace reasons. ... At any rate, you're certainly not changing any > semantics by leaving them all as NullClient, so this patch can't be > more wrong than before. Signed-off-by: Jamey Sharp <jamey@minilop.net> Reviewed-by: Keith Packard <keithp@keithp.com>
Diffstat (limited to 'dix')
-rw-r--r--dix/gc.c46
1 files changed, 3 insertions, 43 deletions
diff --git a/dix/gc.c b/dix/gc.c
index e45378c7b..3a008e0d4 100644
--- a/dix/gc.c
+++ b/dix/gc.c
@@ -91,11 +91,9 @@ ValidateGC(DrawablePtr pDraw, GC *pGC)
* pass NullClient for the client since any access checking should have
* already been done at a higher level.
*
- * Since we had to create a new function anyway, we decided to change the
- * way the list of gc values is passed to eliminate the compiler warnings
- * caused by the DoChangeGC interface. You can pass the values via pC32
- * or pUnion, but not both; one of them must be NULL. If you don't need
- * to pass any pointers, you can use either one:
+ * You can pass the list of gc values via pC32 or pUnion, but not both;
+ * one of them must be NULL. If you don't need to pass any pointers,
+ * you can use either one:
*
* example calling dixChangeGC using pC32 parameter
*
@@ -126,13 +124,6 @@ ValidateGC(DrawablePtr pDraw, GC *pGC)
* Note: we could have gotten by with just the pUnion parameter, but on
* 64 bit machines that would have forced us to copy the value list that
* comes in the ChangeGC request.
- *
- * Ideally, we'd change all the DoChangeGC calls to dixChangeGC, but this
- * is far too many changes to consider at this time, so we've only
- * changed the ones that caused compiler warnings. New code should use
- * dixChangeGC.
- *
- * dpw
*/
#define NEXTVAL(_type, _var) { \
@@ -508,37 +499,6 @@ ChangeGC(GC *pGC, BITS32 mask, XID *pval)
return (dixChangeGC(NullClient, pGC, mask, pval, NULL));
}
-/* DoChangeGC(pGC, mask, pval, fPointer)
- mask is a set of bits indicating which values to change.
- pval contains an appropriate value for each mask.
- fPointer is true if the values for tiles, stipples, fonts or clipmasks
- are pointers instead of IDs. Note: if you are passing pointers you
- MUST declare the array of values as type pointer! Other data types
- may not be large enough to hold pointers on some machines. Yes,
- this means you have to cast to (XID *) when you pass the array to
- DoChangeGC. Similarly, if you are not passing pointers (fPointer = 0) you
- MUST declare the array as type XID (not unsigned long!), or again the wrong
- size data type may be used. To avoid this cruftiness, use dixChangeGC
- above.
-
- if there is an error, the value is marked as changed
- anyway, which is probably wrong, but infrequent.
-
-NOTE:
- all values sent over the protocol for ChangeGC requests are
-32 bits long
-*/
-int
-DoChangeGC(GC *pGC, BITS32 mask, XID *pval, int fPointer)
-{
- if (fPointer)
- /* XXX might be a problem on 64 bit big-endian servers */
- return dixChangeGC(NullClient, pGC, mask, NULL, (ChangeGCValPtr)pval);
- else
- return dixChangeGC(NullClient, pGC, mask, pval, NULL);
-}
-
-
/* CreateGC(pDrawable, mask, pval, pStatus)
creates a default GC for the given drawable, using mask to fill
in any non-default values.