github/acrn-hypervisor
1
0
Fork 0
mirror of https://github.com/projectacrn/acrn-hypervisor.git synced 2025-08-01 16:00:46 +00:00
Code Issues Projects Releases Wiki Activity

vm reset: refine platform reset

Browse Source 
We did following to do platform reset:
1. Try ACPI reset first if it's available
2. Then try 0xcf9 reset method
3. if 2 fails, try keyboard reset method
This introduces some timing concern which needs be handled carefully.

We change it by following:
assume the platforms which ACRN could be run on must support either
ACPI reset or 0xcf9 reset. And simplify platform reset operation
a little bit:
    If ACPI reset register is generated
        try ACPI reset
    else
        try 0xcf9 reset method

Tracked-On: #3609
Signed-off-by: Yin Fengwei <fengwei.yin@intel.com>
This commit is contained in:
Yin Fengwei 2019-08-26 09:42:20 +08:00 committed by ACRN System Integration
parent cd40980d5f
commit 81435f5504
1 changed files with 10 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
hypervisor/arch/x86/pm.c
View File

@ -208,6 +208,11 @@ void reset_host(void)
/* TODO: gracefully shut down all guests before doing host reset. */
/*
* Assumption:
* The platform we are running must support at least one of reset method:
* - ACPI reset
* - 0xcf9 reset
*
* UEFI more likely sets the reset value as 0x6 (not 0xe) for 0xcf9 port.
* This asserts PLTRST# to reset devices on the platform, but not the
* SLP_S3#/4#/5# signals, which power down the systems. This might not be
@ -217,22 +222,13 @@ void reset_host(void)
(gas->bit_width == 8U) && (gas->bit_offset == 0U) &&
(gas->address != 0U) && (gas->address != 0xcf9U)) {
pio_write8(host_reset_reg.val, (uint16_t)host_reset_reg.reg.address);
} else {
/* making sure bit 2 (RST_CPU) is '0', when the reset command is issued. */
pio_write8(0x2U, 0xcf9U);
udelay(50U);
pio_write8(0xeU, 0xcf9U);
}
/*
* Fall back
* making sure bit 2 (RST_CPU) is '0', when the reset command is issued.
*/
pio_write8(0x2U, 0xcf9U);
pio_write8(0xeU, 0xcf9U);
/*
* Fall back
* keyboard controller might cause the INIT# being asserted,
* and not power cycle the system.
*/
pio_write8(0xfeU, 0x64U);
pr_fatal("%s(): can't reset host.", __func__);
while (1) {
asm_pause();