Changeset 76 for trunk/src/os2ahci/ata.h

Timestamp:

Feb 18, 2011, 10:09:10 PM (14 years ago)

Author:

chris

Message:

• APM support
• Generic IOCTL interface for adapter passthrough commands (ATA and ATAPI)
• Fixes to ATAPI sense data handling
• Cosmetic changes to debug reporting
• Fixed missing interrupt enable flag for PIO transfer completions
• Added command line switch /I to ignore specific adapters

File:

• trunk/src/os2ahci/ata.h (modified) (1 diff)

trunk/src/os2ahci/ata.h

@@ -424 +424 @@
 
 /******************************************************************************
- * Generic ATA-8 command structure. This is based on ATA-8, i.e. we won't have
- * to deal with different command structures depending on whether a device is
- * 48-bit capable, etc. because this is handled by the transport (AHCI in our
- * case). Unsupported bits have to be cleared to zero, of course (e.g. the
- * high 20 bits of the LBA when talking to a device that supports only 28
- * bits).
+ * Generic ATA-8 command structure. This is loosely based on ATA-8, i.e. we
+ * don't have to deal with shadow registers and map them to the low bytes of
+ * adjecent words, etc. but there are still some oddities which need to be
+ * taken into consideration. For example, ATA-8 says LBA-28 sector addresses
+ * are simply the lower 28 bits in the LBA field. However, the underlying
+ * transport (SATA in our case) only looks at the three lower bytes of the
+ * LBA field, much like an IDE device would do. This means that only 3 bytes
+ * of the LBA field are processed and this yields only 24 bits. The remaining
+ * 4 bits need to be stored in the "device" register....
+ *
+ * Everything else seems to behave normally, as far as this term can be
+ * applied to IDE/ATA. Further details can be found in section 7.1.5.1 of the
+ * ATA-8 spec (Inputs for 28-bit Read/Write Commands).
  */
 typedef struct {