Reading raw audio sectors

[showka]

Great! Glad to hear it worked.

[Multimedia Mike]

Update on the original query: cdrom_set_sector_size() doesn't work as is (in the current version of KOS). However, I was able to modify it to make it work (modified one of the parameters while reconciling the KOS code with a really ancient DC program that does this correctly).

It's one of those things that makes me wonder if it's supposed to be broken.

[BlueCrab]

I don't think there's any particular reason for it to be broken... I'm guessing this should fix it (as this is how I do it in Yabause, although in assembly, not in C):

Code: Select all

diff --git a/kernel/arch/dreamcast/hardware/cdrom.c b/kernel/arch/dreamcast/hardware/cdrom.c
index 90fefba..6b4e66f 100644
--- a/kernel/arch/dreamcast/hardware/cdrom.c
+++ b/kernel/arch/dreamcast/hardware/cdrom.c
@@ -208,11 +208,19 @@ int cdrom_reinit() {
 
     /* Check disc type and set parameters */
     gdc_get_drv_stat(params);
-    cdxa = params[1] == 32;
-    params[0] = 0;              /* 0 = set, 1 = get */
-    params[1] = 8192;           /* ? */
-    params[2] = cdxa ? 2048 : 1024;     /* CD-XA mode 1/2 */
-    params[3] = sector_size;        /* sector size */
+    if(sector_size != 2352) {
+        cdxa = params[1] == 32;
+        params[0] = 0;              /* 0 = set, 1 = get */
+        params[1] = 8192;           /* ? */
+        params[2] = cdxa ? 2048 : 1024;     /* CD-XA mode 1/2 */
+        params[3] = sector_size;        /* sector size */
+    }
+    else {
+        params[0] = 0;
+        params[1] = 4096;
+        params[2] = 1024;
+        params[3] = 2352;
+    }
 
     if(gdc_change_data_type(params) < 0) {
         rv = ERR_SYS;

[Multimedia Mike]

I don't think there's any particular reason for it to be broken... I'm guessing this should fix it (as this is how I do it in Yabause, although in assembly, not in C):

Yeah, this looks pretty close to what I determined empirically. Do you think you could get it into the official tree?

Thanks.

[BlueCrab]

I don't think there's any particular reason for it to be broken... I'm guessing this should fix it (as this is how I do it in Yabause, although in assembly, not in C):

Yeah, this looks pretty close to what I determined empirically. Do you think you could get it into the official tree?

Thanks.

If you can confirm that the patch I posted works, I'll commit it.

[Quzar]

Blue, why did you change params[1] to be 4096 for 2352 sectors and 8192 otherwise? Do we know what this param does?

The last change to the cdrom code was actually mine that I needed for reading raw sectors from PCE-CD games for HuCast. iirc it worked fine there, but they weren't audio sectors, they were data.

[BlueCrab]

Blue, why did you change params[1] to be 4096 for 2352 sectors and 8192 otherwise? Do we know what this param does?

The last change to the cdrom code was actually mine that I needed for reading raw sectors from PCE-CD games for HuCast. iirc it worked fine there, but they weren't audio sectors, they were data.

I know there was some reason for it explained many moons ago. Basically, it boiled down to 4096 being a "magic" value for 2352 byte sectors. I'm pretty sure I needed it for raw reads in Yabause, so I'm not sure why you wouldn't need it for HuCast.

[Quzar]

Just went back to my sources. Turns out I don't have sources for the last released version, and the one I do have just has notes saying that I can't do the sector reading -_-. I misremembered, the reason for the update I did for KOS was towards that goal, but it seems I never achieved it. This may very well have been why. Hrm.

[T_chan]

Blue, why did you change params[1] to be 4096 for 2352 sectors and 8192 otherwise? Do we know what this param does?

The last change to the cdrom code was actually mine that I needed for reading raw sectors from PCE-CD games for HuCast. iirc it worked fine there, but they weren't audio sectors, they were data.

I know there was some reason for it explained many moons ago. Basically, it boiled down to 4096 being a "magic" value for 2352 byte sectors. I'm pretty sure I needed it for raw reads in Yabause, so I'm not sure why you wouldn't need it for HuCast.

Its indeed a magic value.
I found it many many years ago by doing a little disassembly of the dc bios,
and posted my findings on the KOS mailing list.

see: viewtopic.php?f=29&t=83304

[BlueCrab]

Blue, why did you change params[1] to be 4096 for 2352 sectors and 8192 otherwise? Do we know what this param does?

The last change to the cdrom code was actually mine that I needed for reading raw sectors from PCE-CD games for HuCast. iirc it worked fine there, but they weren't audio sectors, they were data.

I know there was some reason for it explained many moons ago. Basically, it boiled down to 4096 being a "magic" value for 2352 byte sectors. I'm pretty sure I needed it for raw reads in Yabause, so I'm not sure why you wouldn't need it for HuCast.

Its indeed a magic value.
I found it many many years ago by doing a little disassembly of the dc bios,
and posted my findings on the KOS mailing list.

see: viewtopic.php?f=29&t=83304

Aha! I knew I had seen it somewhere before.