Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | README FOR LINUX SONY CDU-535/531 DRIVER |
| 2 | ======================================== |
| 3 | |
| 4 | This is the Sony CDU-535 (and 531) driver version 0.7 for Linux. |
| 5 | I do not think I have the documentation to add features like DMA support |
| 6 | so if anyone else wants to pursue it or help me with it, please do. |
| 7 | (I need to see what was done for the CDU-31A driver -- perhaps I can |
| 8 | steal some of that code.) |
| 9 | |
| 10 | This is a Linux device driver for the Sony CDU-535 CDROM drive. This is |
| 11 | one of the older Sony drives with its own interface card (Sony bus). |
| 12 | The DOS driver for this drive is named SONY_CDU.SYS - when you boot DOS |
| 13 | your drive should be identified as a SONY CDU-535. The driver works |
| 14 | with a CDU-531 also. One user reported that the driver worked on drives |
| 15 | OEM'ed by Procomm, drive and interface board were labelled Procomm. |
| 16 | |
| 17 | The Linux driver is based on Corey Minyard's sonycd 0.3 driver for |
| 18 | the CDU-31A. Ron Jeppesen just changed the commands that were sent |
| 19 | to the drive to correspond to the CDU-535 commands and registers. |
| 20 | There were enough changes to let bugs creep in but it seems to be stable. |
| 21 | Ron was able to tar an entire CDROM (should read all blocks) and built |
| 22 | ghostview and xfig off Walnut Creek's X11R5/GNU CDROM. xcdplayer and |
| 23 | workman work with the driver. Others have used the driver without |
| 24 | problems except those dealing with wait loops (fixed in third release). |
| 25 | Like Minyard's original driver this one uses a polled interface (this |
| 26 | is also the default setup for the DOS driver). It has not been tried |
| 27 | with interrupts or DMA enabled on the board. |
| 28 | |
| 29 | REQUIREMENTS |
| 30 | ============ |
| 31 | |
| 32 | - Sony CDU-535 drive, preferably without interrupts and DMA |
| 33 | enabled on the card. |
| 34 | |
| 35 | - Drive must be set up as unit 1. Only the first unit will be |
| 36 | recognized |
| 37 | |
| 38 | - You must enter your interface address into |
| 39 | /usr/src/linux/drivers/cdrom/sonycd535.h and build the |
| 40 | appropriate kernel or use the "kernel command line" parameter |
| 41 | sonycd535=0x320 |
| 42 | with the correct interface address. |
| 43 | |
| 44 | NOTES: |
| 45 | ====== |
| 46 | |
| 47 | 1) The drive MUST be turned on when booting or it will not be recognized! |
| 48 | (but see comments on modularized version below) |
| 49 | |
| 50 | 2) when the cdrom device is opened the eject button is disabled to keep the |
| 51 | user from ejecting a mounted disk and replacing it with another. |
| 52 | Unfortunately xcdplayer and workman also open the cdrom device so you |
| 53 | have to use the eject button in the software. Keep this in mind if your |
| 54 | cdrom player refuses to give up its disk -- exit workman or xcdplayer, or |
| 55 | umount the drive if it has been mounted. |
| 56 | |
| 57 | THANKS |
| 58 | ====== |
| 59 | |
| 60 | Many thanks to Ron Jeppesen (ronj.an@site007.saic.com) for getting |
| 61 | this project off the ground. He wrote the initial release |
| 62 | and the first two patches to this driver (0.1, 0.2, and 0.3). |
| 63 | Thanks also to Eberhard Moenkeberg (emoenke@gwdg.de) for prodding |
| 64 | me to place this code into the mainstream Linux source tree |
| 65 | (as of Linux version 1.1.91), as well as some patches to make |
| 66 | it a better device citizen. Further thanks to Joel Katz |
| 67 | <joelkatz@webchat.org> for his MODULE patches (see details below), |
| 68 | Porfiri Claudio <C.Porfiri@nisms.tei.ericsson.se> for patches |
| 69 | to make the driver work with the older CDU-510/515 series, and |
| 70 | Heiko Eissfeldt <heiko@colossus.escape.de> for pointing out that |
| 71 | the verify_area() checks were ignoring the results of said checks. |
| 72 | |
| 73 | (Acknowledgments from Ron Jeppesen in the 0.3 release:) |
| 74 | Thanks to Corey Minyard who wrote the original CDU-31A driver on which |
| 75 | this driver is based. Thanks to Ken Pizzini and Bob Blair who provided |
| 76 | patches and feedback on the first release of this driver. |
| 77 | |
| 78 | Ken Pizzini |
| 79 | ken@halcyon.com |
| 80 | |
| 81 | ------------------------------------------------------------------------------ |
| 82 | (The following is from Joel Katz <joelkatz@webchat.org>.) |
| 83 | |
| 84 | To build a version of sony535.o that can be installed as a module, |
| 85 | use the following command: |
| 86 | |
| 87 | gcc -c -D__KERNEL__ -DMODULE -O2 sonycd535.c -o sonycd535.o |
| 88 | |
| 89 | To install the module, simply type: |
| 90 | |
| 91 | insmod sony535.o |
| 92 | or |
| 93 | insmod sony535.o sonycd535=<address> |
| 94 | |
| 95 | And to remove it: |
| 96 | |
| 97 | rmmod sony535 |
| 98 | |
| 99 | The code checks to see if MODULE is defined and behaves as it used |
| 100 | to if MODULE is not defined. That means your patched file should behave |
| 101 | exactly as it used to if compiled into the kernel. |
| 102 | |
| 103 | I have an external drive, and I usually leave it powered off. I used |
| 104 | to have to reboot if I needed to use the CDROM drive. Now I don't. |
| 105 | |
| 106 | Even if you have an internal drive, why waste the 96K of memory |
| 107 | (unswappable) that the driver uses if you use your CD-ROM drive infrequently? |
| 108 | |
| 109 | This driver will not install (whether compiled in or loaded as a |
| 110 | module) if the CDROM drive is not available during its initialization. This |
| 111 | means that you can have the driver compiled into the kernel and still load |
| 112 | the module later (assuming the driver doesn't install itself during |
| 113 | power-on). This only wastes 12K when you boot with the CDROM drive off. |
| 114 | |
| 115 | This is what I usually do; I leave the driver compiled into the |
| 116 | kernel, but load it as a module if I powered the system up with the drive |
| 117 | off and then later decided to use the CDROM drive. |
| 118 | |
| 119 | Since the driver only uses a single page to point to the chunks, |
| 120 | attempting to set the buffer cache to more than 2 Megabytes would be very |
| 121 | bad; don't do that. |