diff options
author | Kartik Agaram <vc@akkartik.com> | 2021-01-24 11:15:09 -0800 |
---|---|---|
committer | Kartik Agaram <vc@akkartik.com> | 2021-01-24 19:43:23 -0800 |
commit | 09f2f3075a8be46484be2b5900888a8ac4799b75 (patch) | |
tree | 595981fd98a11452d4b664afdaff83316df87d39 | |
parent | dba18f7f6f07840bf981d21a9acabd33f651c860 (diff) | |
download | mu-09f2f3075a8be46484be2b5900888a8ac4799b75.tar.gz |
7555 - snapshot
While baremetal has been working with Qemu, it's been broken with Bochs since commit 7547, where we started reading more than 63 sectors (1 track) from disk. Good to know that Bochs simulates native hardware with so much verisimilitude! Unfortunately things aren't fixed yet. The current state: - Qemu - - Bochs - ex2.hex never switches modes works ex2.subx never switches modes works ex2.mu never switches modes fails unit tests It sucks that Bochs doesn't have strictly superior verisimilitude compared to Qemu :(
-rw-r--r-- | baremetal/boot.hex | 43 |
1 files changed, 30 insertions, 13 deletions
diff --git a/baremetal/boot.hex b/baremetal/boot.hex index 50fdb334..c05a3f19 100644 --- a/baremetal/boot.hex +++ b/baremetal/boot.hex @@ -74,21 +74,43 @@ # calls, so we don't need to initialize the stack. # 0e: - # load some sectors from disk + # load 62 sectors from disk into addresses [0x7e00, 0xfa00) b4 02 # ah <- 2 # read sectors from disk +# 10: # dl comes conveniently initialized at boot time with the index of the device being booted b5 00 # ch <- 0 # cylinder 0 b6 00 # dh <- 0 # track 0 b1 02 # cl <- 2 # second sector, 1-based - b0 80 # al <- 128 # number of sectors to read; TODO - all sectors might need to be in a single track on real hardware (so 63 sectors at most including the boot sector) + b0 3e # al <- 62 # number of sectors to read # address to write sectors to = es:bx = 0x7e00, contiguous with boot segment bb 00 00 # bx <- 0 8e c3 # es <- bx bb 00 7e # bx <- 0x7e00 [label] +# 20: cd 13 # int 13h, BIOS disk service - 0f 82 8a 00 # jump-if-carry disk-error [label] - + 0f 82 8a 00 # jump-if-carry disk_error [label] # 26: + # load 63 sectors from disk into addresses [0xfa00, 0x17800) + b4 02 # ah <- 2 # read sectors from disk + b5 00 # ch <- 0 # cylinder 0 + b6 01 # dh <- 1 # track 1 + b1 01 # cl <- 1 # first sector, 1-based + b0 3f # al <- 63 # number of sectors to read +# 30: + # "Addressing of Buffer should guarantee that the complete buffer is inside + # the given segment, i.e. ( BX + size_of_buffer ) <= 10000h." + # https://en.wikipedia.org/wiki/INT_13H#INT_13h_AH=02h:_Read_Sectors_From_Drive + bb 00 fa # bx <- 0xfa00 [label] + 8e c3 # es <- bx + bb 00 00 # bx <- 0 + cd 13 # int 13h, BIOS disk service + 0f 82 72 00 # jump-if-carry disk_error [label] +# 3e: + # reset es + bb 00 00 # bx <- 0 + 8e c3 # es <- bx + +# 43: # undo the A20 hack: https://en.wikipedia.org/wiki/A20_line # this is from https://github.com/mit-pdos/xv6-public/blob/master/bootasm.S # seta20.1: @@ -99,7 +121,7 @@ b0 d1 # al <- 0xd1 e6 64 # port 0x64 <- al -# 30: +# 4d: # seta20.2: e4 64 # al <- port 0x64 a8 02 # set zf if bit 1 (second-least significant) is not set @@ -108,7 +130,7 @@ b0 df # al <- 0xdf e6 64 # port 0x64 <- al -# 3a: +# 57: # adjust video mode b4 4f # ah <- 4f (VBE) b0 02 # al <- 02 (set video mode) @@ -117,7 +139,7 @@ # fallback mode: 0x0101 (640x480x256) cd 10 # int 10h, Vesa BIOS extensions -# 43: +# 60: # load information for the (hopefully) current video mode # mostly just for the address to the linear frame buffer b4 4f # ah <- 4f (VBE) @@ -126,7 +148,7 @@ bf 00 7f # di <- 0x7f00 (video mode info) [label] cd 10 -# 4f: +# 6c: # switch to 32-bit mode 0f 01 16 # lgdt 00/mod/indirect 010/subop 110/rm/use-disp16 a0 7c # *gdt_descriptor [label] @@ -135,11 +157,6 @@ 0f 22 c0 # cr0 <- eax ea e0 7c 08 00 # far jump to initialize_32bit_mode after setting cs to the record at offset 8 in the gdt (gdt_code) [label] -# padding -# 63: - 00 00 00 00 00 00 00 00 00 00 00 00 00 -00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 - ## GDT: 3 records of 8 bytes each # 80: |