Commit a676f8d0 authored by Peter Osterlund's avatar Peter Osterlund Committed by Linus Torvalds
Browse files

[PATCH] pktcdvd: documentation update

Update the "theory of operation" description.
Signed-off-by: default avatarPeter Osterlund <>
Signed-off-by: default avatarAndrew Morton <>
Signed-off-by: default avatarLinus Torvalds <>
parent d0272e78
......@@ -5,29 +5,41 @@
* May be copied or modified under the terms of the GNU General Public
* License. See linux/COPYING for more information.
* Packet writing layer for ATAPI and SCSI CD-R, CD-RW, DVD-R, and
* DVD-RW devices (aka an exercise in block layer masturbation)
* Packet writing layer for ATAPI and SCSI CD-RW, DVD+RW, DVD-RW and
* DVD-RAM devices.
* Theory of operation:
* TODO: (circa order of when I will fix it)
* - Only able to write on CD-RW media right now.
* - check host application code on media and set it in write page
* - interface for UDF <-> packet to negotiate a new location when a write
* fails.
* - handle OPC, especially for -RW media
* At the lowest level, there is the standard driver for the CD/DVD device,
* typically ide-cd.c or sr.c. This driver can handle read and write requests,
* but it doesn't know anything about the special restrictions that apply to
* packet writing. One restriction is that write requests must be aligned to
* packet boundaries on the physical media, and the size of a write request
* must be equal to the packet size. Another restriction is that a
* GPCMD_FLUSH_CACHE command has to be issued to the drive before a read
* command, if the previous command was a write.
* Theory of operation:
* The purpose of the packet writing driver is to hide these restrictions from
* higher layers, such as file systems, and present a block device that can be
* randomly read and written using 2kB-sized blocks.
* The lowest layer in the packet writing driver is the packet I/O scheduler.
* Its data is defined by the struct packet_iosched and includes two bio
* queues with pending read and write requests. These queues are processed
* by the pkt_iosched_process_queue() function. The write requests in this
* queue are already properly aligned and sized. This layer is responsible for
* issuing the flush cache commands and scheduling the I/O in a good order.
* We use a custom make_request_fn function that forwards reads directly to
* the underlying CD device. Write requests are either attached directly to
* a live packet_data object, or simply stored sequentially in a list for
* later processing by the kcdrwd kernel thread. This driver doesn't use
* any elevator functionally as defined by the elevator_s struct, but the
* underlying CD device uses a standard elevator.
* The next layer transforms unaligned write requests to aligned writes. This
* transformation requires reading missing pieces of data from the underlying
* block device, assembling the pieces to full packets and queuing them to the
* packet I/O scheduler.
* This strategy makes it possible to do very late merging of IO requests.
* A new bio sent to pkt_make_request can be merged with a live packet_data
* object even if the object is in the data gathering state.
* At the top layer there is a custom make_request_fn function that forwards
* read requests directly to the iosched queue and puts write requests in the
* unaligned write queue. A kernel thread performs the necessary read
* gathering to convert the unaligned writes to aligned writes and then feeds
* them to the packet I/O scheduler.
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment