[PATCH 2/4 v2] fbdev: export fb_edid_add_monspecs() for modules,

fb_edid_add_monspecs() should also be exported for use in modules, and it
requires a dummy version for the case, when CONFIG_FB_MODE_HELPERS is not
selected. This patch also improves the algorithm by removing a redundant
memory allocation, adds function documentation, adds data verification and
replaces memmove() with memcpy().

Signed-off-by: Guennadi Liakhovetski <redacted>
---

v2: added changes, suggested by Andrew:

1. replace memmove() with memcpy()
2. add documentation
3. sanity check edid[2]

 drivers/video/fbmon.c |   58 +++++++++++++++++++++++++-----------------------
 1 files changed, 30 insertions(+), 28 deletions(-)

diff --git a/drivers/video/fbmon.c b/drivers/video/fbmon.c
index a0b5a93..38b3c30 100644
--- a/drivers/video/fbmon.c
+++ b/drivers/video/fbmon.c

@@ -973,58 +973,56 @@ void fb_edid_to_monspecs(unsigned char *edid, struct fb_monspecs *specs)
 	DPRINTK("====================\n");
 }
 
+/**
+ * fb_edid_add_monspecs() - add monitor video modes from E-EDID data
+ * @edid:	128 byte array with an E-EDID block
+ * @spacs:	monitor specs to be extended
+ */
 void fb_edid_add_monspecs(unsigned char *edid, struct fb_monspecs *specs)
 {
 	unsigned char *block;
-	struct fb_videomode *mode, *m;
-	int num = 0, i, first = 1;
+	struct fb_videomode *m;
+	int num = 0, i;
+	u8 edt[(128 - 4) / DETAILED_TIMING_DESCRIPTION_SIZE];
 
-	if (edid = NULL)
+	if (!edid)
 		return;
 
 	if (!edid_checksum(edid))
 		return;
 
-	if (edid[0] != 0x2)
+	if (edid[0] != 0x2 ||
+	    edid[2] < 4 || edid[2] > 128 - DETAILED_TIMING_DESCRIPTION_SIZE)
 		return;
 
-	mode = kzalloc(50 * sizeof(struct fb_videomode), GFP_KERNEL);
-	if (mode = NULL)
-		return;
-
-	block = edid + edid[0x2];
+	block = edid + edid[2];
 
 	DPRINTK("  Extended Detailed Timings\n");
 
-	for (i = 0; i < (128 - edid[0x2]) / DETAILED_TIMING_DESCRIPTION_SIZE;
-	     i++, block += DETAILED_TIMING_DESCRIPTION_SIZE) {
-		if (!(block[0] = 0x00 && block[1] = 0x00)) {
-			get_detailed_timing(block, &mode[num]);
-			if (first) {
-			        mode[num].flag |= FB_MODE_IS_FIRST;
-				first = 0;
-			}
-			num++;
-		}
-	}
+	for (i = 0; i < (128 - edid[2]) / DETAILED_TIMING_DESCRIPTION_SIZE;
+	     i++, block += DETAILED_TIMING_DESCRIPTION_SIZE)
+		if (!(block[0] = 0x00 && block[1] = 0x00))
+			edt[num++] = block - edid;
 
 	/* Yikes, EDID data is totally useless */
-	if (!num) {
-		kfree(mode);
+	if (!num)
 		return;
-	}
 
 	m = kzalloc((specs->modedb_len + num) *
 		       sizeof(struct fb_videomode), GFP_KERNEL);
 
-	if (!m) {
-		kfree(mode);
+	if (!m)
 		return;
+
+	memcpy(m, specs->modedb, specs->modedb_len * sizeof(struct fb_videomode));
+
+	for (i = specs->modedb_len; i < specs->modedb_len + num; i++) {
+		get_detailed_timing(edid + edt[i - specs->modedb_len], &m[i]);
+		if (i = specs->modedb_len)
+			m[i].flag |= FB_MODE_IS_FIRST;
+		pr_debug("Adding %ux%u@%u\n", m[i].xres, m[i].yres, m[i].refresh);
 	}
 
-	memmove(m, specs->modedb, specs->modedb_len * sizeof(struct fb_videomode));
-	memmove(m + specs->modedb_len, mode, num * sizeof(struct fb_videomode));
-	kfree(mode);
 	kfree(specs->modedb);
 	specs->modedb = m;
 	specs->modedb_len = specs->modedb_len + num;

@@ -1346,6 +1344,9 @@ void fb_edid_to_monspecs(unsigned char *edid, struct fb_monspecs *specs)
 {
 	specs = NULL;
 }
+void fb_edid_add_monspecs(unsigned char *edid, struct fb_monspecs *specs)
+{
+}
 void fb_destroy_modedb(struct fb_videomode *modedb)
 {
 }

@@ -1453,6 +1454,7 @@ EXPORT_SYMBOL(fb_firmware_edid);
 
 EXPORT_SYMBOL(fb_parse_edid);
 EXPORT_SYMBOL(fb_edid_to_monspecs);
+EXPORT_SYMBOL(fb_edid_add_monspecs);
 EXPORT_SYMBOL(fb_get_mode);
 EXPORT_SYMBOL(fb_validate_mode);
 EXPORT_SYMBOL(fb_destroy_modedb);

-- 
1.7.2.3

On Tue, Nov 02, 2010 at 11:47:36AM +0100, Guennadi Liakhovetski wrote:

[snip]

This can obviously fail, and should be returning the error value instead
of simply quashing it.

On Tue, Nov 02, 2010 at 11:47:36AM +0100, Guennadi Liakhovetski wrote:

...

+	u8 edt[(128 - 4) / DETAILED_TIMING_DESCRIPTION_SIZE];
 

...

...

+	block = edid + edid[2];
 

...

Can we also do something about these magic numbers? There is already a
drivers/video/edid.h that you can fill out as necessary. I realize that
these values are probably pretty obvious if you have a copy of the spec
in front of you, I don't.

On Wed, 10 Nov 2010, Paul Mundt wrote:

On Tue, Nov 02, 2010 at 11:47:36AM +0100, Guennadi Liakhovetski wrote:

quoted

diff --git a/drivers/video/fbmon.c b/drivers/video/fbmon.c
index a0b5a93..38b3c30 100644
--- a/drivers/video/fbmon.c
+++ b/drivers/video/fbmon.c

@@ -973,58 +973,56 @@ void fb_edid_to_monspecs(unsigned char *edid, struct fb_monspecs *specs)
 	DPRINTK("====================\n");
 }
 
+/**
+ * fb_edid_add_monspecs() - add monitor video modes from E-EDID data
+ * @edid:	128 byte array with an E-EDID block
+ * @spacs:	monitor specs to be extended
+ */
 void fb_edid_add_monspecs(unsigned char *edid, struct fb_monspecs *specs)
 {

[snip]

quoted

 	m = kzalloc((specs->modedb_len + num) *
 		       sizeof(struct fb_videomode), GFP_KERNEL);
 
-	if (!m) {
-		kfree(mode);
+	if (!m)
 		return;
+

This can obviously fail, and should be returning the error value instead
of simply quashing it.

Yes, this can fail, just as well as one of the other tests in the 
beginning of this function:

	if (!edid)
		return;

	if (!edid_checksum(edid))
		return;

	if (edid[0] != 0x2 ||
	    edid[2] < 4 || edid[2] > 128 - DETAILED_TIMING_DESCRIPTION_SIZE)
		return;

The idea behind the present behaviour is: the function is trying to add 
more modes to the existing modedb, but it can also fail for a number of 
reasons (e.g., no modes found in the new EDID block), in which case you 
just keep your existing modedb. So, in principle, you are not very much 
interested in whether this function managed to add any new modes or not - 
you just use whatever is there. Would you really prefer to change this?

Thanks
Guennadi
---
Guennadi Liakhovetski, Ph.D.
Freelance Open-Source Software Developer
http://www.open-technology.de/