Since yesterday's patch, sr and sg appear to have been fixed, but the SCSI
tape driver still has kdev_t problems. Having fat-fingered my way into a
disk disaster yesterday, I have a real interest in a working tape drive :)
Here's a rework of my st.c kdev_t patch for 2.5.2-pre7.
jon
Jonathan Corbet
Executive editor, LWN.net
[email protected]
diff -r -u 2.5.2-pre6/drivers/scsi/st.c linux-2.5.2-pre6-scsi/drivers/scsi/st.c
--- 2.5.2-pre6/drivers/scsi/st.c Thu Jan 3 11:35:26 2002
+++ linux-2.5.2-pre6-scsi/drivers/scsi/st.c Thu Jan 3 11:38:22 2002
@@ -133,8 +133,8 @@
#define ST_TIMEOUT (900 * HZ)
#define ST_LONG_TIMEOUT (14000 * HZ)
-#define TAPE_NR(x) (MINOR(x) & ~(128 | ST_MODE_MASK))
-#define TAPE_MODE(x) ((MINOR(x) & ST_MODE_MASK) >> ST_MODE_SHIFT)
+#define TAPE_NR(x) (minor(x) & ~(128 | ST_MODE_MASK))
+#define TAPE_MODE(x) ((minor(x) & ST_MODE_MASK) >> ST_MODE_SHIFT)
/* Internal ioctl to set both density (uppermost 8 bits) and blocksize (lower
24 bits) */
@@ -878,7 +878,7 @@
}
STp->in_use = 1;
write_unlock_irqrestore(&st_dev_arr_lock, flags);
- STp->rew_at_close = STp->autorew_dev = (MINOR(inode->i_rdev) & 0x80) == 0;
+ STp->rew_at_close = STp->autorew_dev = (minor(inode->i_rdev) & 0x80) == 0;
if (STp->device->host->hostt->module)
__MOD_INC_USE_COUNT(STp->device->host->hostt->module);
@@ -3717,7 +3717,7 @@
tpnt->tape_type = MT_ISSCSI2;
tpnt->inited = 0;
- tpnt->devt = MKDEV(SCSI_TAPE_MAJOR, i);
+ tpnt->devt = mk_kdev(SCSI_TAPE_MAJOR, i);
tpnt->dirty = 0;
tpnt->in_use = 0;
tpnt->drv_buffer = 1; /* Try buffering if no mode sense */