Skip to contentSkip to conversation

Aureal VOrtex update and MPU401 survey

DISCUSSION Manuel Jander 2 participants 2 messages Mar 13, 2003

lore
Share & tools
lore
#0 lore raw

Hi everyone.

Aurela Vortex:
I'm pleased to announce that the Vortex driver PCM is working. Its not 100% perfect yet, sometimes it skips, but that will be done in the next week (i'll try to add Scatter Gather support).

MPU401 survey:
I though about a modification for the mpu401_uart driver, to allow
a more versatil handling of different IO types. I observed that
at least the SBLive, appart from the Vortex have a standart MPU401 port, but the hardware IO is slighdely different enough to prevent the usage of the existing MPU401 driver. That means the driver is copy pasted and only the Register access (inb and outb) calls are modified.

Know there are too options: allow replacement functions for inb and outb, or to use wrappers that accept the mpu type as parameter so that a more sophisticated IO handling can be done. THat means:

case 1

typedef mpu401_t {
....
void (*write) (u8 data, u32 addr);
u8 (*write) (u32 addr);
....
}
.....

mpu->write = outb;
mpu->read = inb;

----------- < OR >-------------------
case 2

typedef mpu401_t {
....
void (*write) (mpu401_t *mpu, u8 data, u32 addr); u8 (*write) (mpu401_t *mpu, u32 addr);
....
}

...

mpu->write = mpu_write;
mpu->read = mpu_read;

... and implemented in the specific card driver:

void mpu_write(mpu401_t *mpu, u8 data, u32 addr) {
    some_iohardware_write(mpu->private_data, data, addr);
}
void mpu_write(mpu401_t *mpu, u8 data, u32 addr) {
    return some_iohardware_read(mpu->private_data, addr);
}

The question is: Can anyone take advantage of having the "mpu" object avaiable inside a IO wrapper ? The SBLive should, but are the maintainers of the SBLive in the mood to actually make use of that ?

-If the mpu401_uart driver assigns per default the apropiate functions to mpu->write and mpu->read, -no change in the API is needed-. -At least case 1 is needed to avoid having to copy paste the driver for the Aureal Vortex since it uses mmio instead of port io.

Best Regards

Manuel Jander.

-------------------------------------------------------
This SF.net email is sponsored by:Crypto Challenge is now open! Get cracking and register here for some mind boggling fun and the chance of winning an Apple iPod:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en

#1 lore raw

Hi,

At Thu, 13 Mar 2003 00:28:39 -0400,

Quoted 45 lines · Manuel Jander

MPU401 survey:
I though about a modification for the mpu401_uart driver, to allow a more versatil handling of different IO types. I observed that at least the SBLive, appart from the Vortex have a standart MPU401 port, but the hardware IO is slighdely different enough to prevent the usage of the existing MPU401 driver. That means the driver is copy pasted and only the Register access (inb and outb) calls are modified.

Know there are too options: allow replacement functions for inb and outb, or to use wrappers that accept the mpu type as parameter so that a more sophisticated IO handling can be done. THat means:

case 1

typedef mpu401_t {
   ....
   void (*write) (u8 data, u32 addr);
   u8 (*write) (u32 addr);
   ....
}

.....

mpu->write = outb;
mpu->read = inb;

----------- < OR >-------------------
case 2

typedef mpu401_t {
   ....
   void (*write) (mpu401_t *mpu, u8 data, u32 addr);
   u8 (*write) (mpu401_t *mpu, u32 addr);
   ....
}

...

mpu->write = mpu_write;
mpu->read = mpu_read;

... and implemented in the specific card driver:

Quoted 7 lines
void mpu_write(mpu401_t *mpu, u8 data, u32 addr) {
     some_iohardware_write(mpu->private_data, data, addr);
}
void mpu_write(mpu401_t *mpu, u8 data, u32 addr) {
     return some_iohardware_read(mpu->private_data, addr);
}

(i suppose the second one corresponds to read :)

i like the idea. not only emu10k1 but some also other drivers use a copy of mpu401_uart.c code. for compatibility, we can keep the old snd_mpu_uart401_new() function which works like it has been, and add a new function with the driver-specific callbacks.

from the suggestions, i would choose the last one. the callback likely needs a private data to access the hardware properly. also, for the callback argument, we don't need to pass the exact address but only a enum CMD and DATA.

Quoted 3 lines

The question is: Can anyone take advantage of having the "mpu" object avaiable inside a IO wrapper ? The SBLive should, but are the maintainers of the SBLive in the mood to actually make use of that ?

no problem, but unfortunately i (and perhaps Jaroslav, too) have no time in this week to attend the LAD meeting at Karlsruhe. so, the work will be done in the next week.

the only remaining problem is the interrupt handler.
in the case of mpu401_uart.c, there is only one interrupt handler which handles both receive and transmit. but some hardwares have different rx and tx interrupts, so we need to separate them.

anyway, please feel free to submit the patch.

ciao,

Takashi

-------------------------------------------------------
This SF.net email is sponsored by:Crypto Challenge is now open! Get cracking and register here for some mind boggling fun and the chance of winning an Apple iPod:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en