| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
| |
The E820 map ended the usable region at a hardcoded 0x08000000, so a guest
with more RAM had the extra memory reported as reserved and ran out of memory
during initramfs unpacking.
Add cmos_read8() and size the extended memory entry from CMOS registers 0x34
and 0x35, falling back to the 16 MiB units in 0x30 and 0x31 and then to the
previous 128 MiB end when neither is programmed.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
POST only unmasked IRQ1 in the PIC and never configured the 8042 itself, so
the keyboard stayed in its post-reset disabled state: the clock was inhibited
and no scan code ever reached the output buffer. IRQ1 never fired, the BDA
keyboard buffer stayed empty, and GRUB polling INT 16h AH=01h never saw a key.
Add kbd_init() to disable both ports, drain the output buffer, clear the clock
inhibit bits in the command byte, enable the keyboard and auxiliary IRQs and
select set 1 translation, then enable scanning with 0xF4.
The IRQ1 handler now checks the status register for a pending output byte
before reading the data port and drops auxiliary port bytes. The set 2 break
prefix tracking is gone because translation is enabled.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
| |
Signed-off-by: Richard Qin <richard.qin.x@gmail.com>
|
| |
|
|
| |
Signed-off-by: Richard Qin <richard.qin.x@gmail.com>
|
| |
|
|
| |
Signed-off-by: Richard Qin <richard.qin.x@gmail.com>
|
| |
|
|
| |
Signed-off-by: Richard Qin <richard.qin.x@gmail.com>
|
| |
|
|
| |
Signed-off-by: Richard Qin <richard.qin.x@gmail.com>
|
|
|
Signed-off-by: Richard Qin <richard.qin.x@gmail.com>
|
