H8 improvements: H8/3437, I2C master and Sun Fire V120 stub #13016
Conversation
This chip is used as a management controller on some Sun Servers (Fire V120, Netra 120) and IBM Laptops (ThinkPad 600) and perhaps elsewhere. It's very much like a 3337, except it got two extra GPIO ports.
Only master side implemented. Tested against i2cmem and i2chle. Shares register space with the serial port, with a bit that arbitrates between the two.
Just the management microcomputer (Lights-Out Management) with its configuration ROM for now. No main processor.
|
|
| m_ram_start(start) | ||
| { | ||
| } | ||
|
|
||
| h83337_device::h83337_device(const machine_config &mconfig, device_type type, const char *tag, device_t *owner, u32 clock, u32 start) : | ||
| h83337_device(mconfig, H83337, tag, owner, clock, address_map_constructor(FUNC(h83337_device::map), this), 0xf780) |
There was a problem hiding this comment.
The H83337 here should be "type". That breaks mame -valid.
| m_icdr = data; | ||
| } | ||
|
|
||
| u8 h8_i2c_device::icdr_r() |
There was a problem hiding this comment.
That method is iffy. I know it's going to work, but it pretends transmission is instantaneous which is very untrue. We can use it as a starting point, but it should respect the real timings.
| @@ -0,0 +1,57 @@ | |||
| // license:BSD-3-Clause | |||
| // copyright-holders:Olivier Galibert | |||
| // copyright-holders:Lubomir Rintel | |||
There was a problem hiding this comment.
Multiple copyright-holders should be comma-separated.
| @@ -0,0 +1,159 @@ | |||
| #include "emu.h" | |||
There was a problem hiding this comment.
This file is missing the license and copyright-holders headers. You should also put some basic information about the driver at the top.
|
|
||
| namespace { | ||
|
|
||
| class v120_state : public driver_device |
There was a problem hiding this comment.
v120 is relatively generic - maybe it would be better as sunv120?
| m_ttya->set_option_device_input_defaults("terminal", DEVICE_INPUT_DEFAULTS_NAME(terminal)); | ||
|
|
||
| // Debug chatter | ||
| m_lom->scl_cb().set([this](int state) { logerror("scl_write %d\n", state); }); |
There was a problem hiding this comment.
The usual convention is to group related device functions. In this case everything handling m_lom might be better put above where the CPU is configured.
| * ... | ||
| * lom> rw f77e | ||
| * | ||
| * The configuation EEPROM describes which fans, supplies, etc. are |
| } // anonymous namespace | ||
|
|
||
| // YEAR, NAME, PARENT, COMPAT, MACHINE, INPUT, CLASS, INIT, COMPANY, FULLNAME, FLAGS | ||
| COMP( 1996, v120, 0, 0, v120, 0, v120_state, empty_init, "Sun Microsystems", "v120", MACHINE_IS_SKELETON ) |
There was a problem hiding this comment.
The fullname would be Fire V120 or Fire/Netra V120 I suppose?
Hi,
I found MAME and its H8 emulation very useful in my effort to understand management controller on mid 2000's Sun Fire V120 server. I'm wondering if anything I've done is useful for MAME:
Here's what I got:
These three are probably useful for the Thinkpad 600.
Not sure if useful. Very minimal skeleton, since I've only been interested in the management microcomputer.
The I2C implementation has also been tested with a skeleton of a fan monitor chip using the i2chle interface, but that is certainly not something that's ready: https://github.com/lkundrak/mame/commits/lr/lomlite-dev/