为什么磁盘图标有\"PC\"字样?

OS: OS9
设备:U盘

问题: 用系统自带的驱动,磁盘图标没有\"PC\"字样,用USB DDK的例子编译的驱动,磁盘图标有\"PC\"字样.是什么原因呢?

谢了.

MacWin,你的联系方式是什么？给你写了信收到没有？

联系我,请加MSN: xinjian_chen@hotmail.com

DDK里的图标都包含在源文件里，要修改图标，把UnitTableDriverIcons.c里的图标数据改一下，应该就可以了

Disk icon on OS 9


dear Chen,

This message will help you with your recent questions.

I will discuss two topics : hints on driver calls you asked about, and
secondly a list of serious bugs in apples usb source code you need to know
about.


There is a mandatory Apple status call as of its October 20, 1994 draft
rules : cs code 43, DRIVER_GESTALT and extending it there is a selector for
color icons used in the finder (it will need to be cached in ram, because
you cannot access resources during synchronous IO, so copy the color icon
resources into a locked handle area in system heap location.

you need to implement \'mics\' and \'dics\'

dics : returns a pointer to a IconFamily (\'icns\') data structure for
physical drive Disk Driver (formerly in csCode 21) in driverGestaltResponse

\'mics\'is same as \'dics\' but formerly in csCode 22. this is meidia related
color icon.

There are bugs in some versions of apples finder :

one major defect : does not allow you to clear undefined bytes in csParam
field to zero, in many versions of apple finder, just leave undefined bytes
alone because apple uses the bytes illegally in its cd-rom driver usage.

There are other apple bug to worry about.

You might want to check your list of gestalt calls, one is for media type
and can return \'cdrm\' and \'disk\'.

helpful rare gestalt call hints :

kdgVMOptions , set correctly for SPEED, and make driver a ptr not hdl.

you  might want to simulate every feature of apples driver and all its
gestalt responses in order to get the driver to do anything apples drivers
do, including using the gestalt version of apples driver.

if you support luns remember to support deviceident gestalt call so that the
lun can be determined.

I see you are trying to use usb mass storage device driver. the source code
written by Apple\'s usb driver team is unusable unless you know how to locate
and fix all the many bugs in the code. Amny of the bugs shipped in apples
own drivers. Its horribly damaged code.

To be nice, and to document a *FEW* of the many bugs in apples USB mass
storage driver I will list *some* of them in this document to save you a lot
of sad months of lost work. Apple does not pay enough salary to get the
right people in the driver team (I know), and also has limits on contractor
hourly $ rates, so that is how they end up with many of these bugs  :

USB BUG 1: in structure : \"StorageClassInfo\" used to hold Globals

You need to add some safety space after \"Configuration block\" field \"USBPB
usbPB\" 96 bytes are corrupted following the previous \"usbPB\" structure, USB
1.3 and earlier, the call to USBResetDevice would overwrite memory up to 96
bytes past the end of the \"usbPB\" parameter block damaging other vital
globals in \"StorageClassInfo\". you need to ensure your compiler does not
reorganize things on you or create a nested type. In 1.4 of USB Apple Fixed
USBResetDevice so that it no longer corrupts memory past the parameter
block, but I knew about the bug before apple fixed it. The bug is still in
that code though that you are using I bet.


USB BUG 2 : in call StorageClassDriverControl() with selector of
kUSBStorageControlSetParentsRefNumber, apple\'s code returns an error with
unititilized value for \"kUSBStorageControlSetParentsRefNumber\" status.
\"status\" was random off stack. To fix it set \"status = noErr\" then only
knock down if \"theDeviceRef != 0\"


USB BUG 3 : In DriverControlCmd () Apple\'s buggy driver regularly is BLINDLY
clearing the gTheDriverGlobals.drive_Internal_PB on unsupported control
calls EVEN WHEN PARAM BLOCK BUSY!! when selector of kInitializeDeviceAccess
or kTerminateDeviceAccess is called.
gTheDriverGlobals.drive_Internal_PB.ioCommandID  was getting zapped early.
Perhaps the follwing hint may be of partial help :

            err = HandleControlRequest((UInt32)
&gTheDriverGlobals.drive_Internal_PB, pbPtr, &InitializeDriveAccessDone);
            if (err != kRequestPending )
            {
                gTheDriverGlobals.drive_Internal_PB.theIOPB            =
nil;
                gTheDriverGlobals.drive_Internal_PB.ioCommandID         = 0;
            }


USB BUG 4 : APPLE HAS A SERIOUS BUG AND NEEDED ANOTHER SPARE INTERNAL
PARAMETER BLOCK. Otherwise control commands and other commands during
interrupts caused havoc and wrote to Nil. The apple  driver code did not
check for write to nils and used address 0 for structure storage.
Specifically, you need a scond block for ManualEjectPB, and add state logic.


USB BUG 5 : uninitialized variable. in call ShimOpenDriver() called by
notification routine to load a Unit Table driver at the start of the routine
set :
    hDrvrResource = nil;



USB BUG 6: Apple\'s code was resetting too many globals in a loop during
removal deadlocks. So create global knock down flag to avoid unneeded work
and then your \"StorageClassDriverInitialize()\" call can be a nop and return
noErr if the global flag is set. I hate global flags but you need them at
times.


USB BUG 7 : In \"StorageDeviceInitiateConfiguration ()\" add auto repair code
to \"kStorageConfigureInterface\" . usbReference Interface reference obtained
from USBNewInterfaceRef as per required documentation rules.

if (usbPB->usbReference != gStorageClassInfo.interface_Ref) {
    usbPB->usbReference = gStorageClassInfo.interface_Ref;
}

USB BUG 8 : InstallPB structure was not ansi C compliant. use the following
:
typedef struct
{
    volatile OSStatus                status;
    void                             *ourCompletion;
    UInt32                            capacity;
    UInt32                            blockSize;
    OSType                            currentMediaType;
    UInt8                            currentState;
    UInt32                            currentPartBlock;
    UInt32                            lastPartBlock;
    DriveInstallCompletionProcPtr    installCompletion;
    Boolean                            useNativeBlocks;
    UInt8                            buffer[2048];
}InstallPB, *InstallPBPtr;


USB BUG 9 : DriverGestaltDeviceModelInfoResponse Apple had a screwed up
typedef. use :

typedef struct {
  UInt32              infoStructVersion;
  StringPtr           vendorName;
  StringPtr           productName;
  StringPtr           revisionNumber;
  StringPtr           subRevisionNumber;
  StringPtr           serialNumber;
} DriverGestaltDeviceModelInfoResponse;



I stopped keeping track of usb bugs I fixed and located in apples example
mass storage class driver code, and ended up rewriting it all heavily. I
also have not revealed all the apple usb bugs I know about deliberately. I
was just trying to make a point.

Apple\'s USB architecture is polled and slow and cannot handle a synch IO
call from another storage driver during interrupts. Its more like a toy that
emulates \"network\" storage. Firewire si no better. ATA IDE and SCSI work
flawlessly on macs and have immediate IO, and even IOs while interrupts off.

I am warning you though that you would be better off trying to write a
perfect SCSI style DVD CD driver and then do the minimal low level support
to make it work with minimal but proper USB code to function  with USB
devices of your choosing.

I can be hired to fix driver issues on a variety of OSes.

good luck, chen

you will need a lot of luck, or a lot of skill.

Ling Chao