Commit Graph

157 Commits (acf572905c53d7a057203e3fbc1fd81606c0f375)

Author SHA1 Message Date
Michael Brown d48d0fb1bb Add the concept of a "user pointer" (similar to the void __user * in
the kernel), which encapsulates the information needed to refer to an
external buffer.  Under normal operation, this can just be a void *
equivalent, but under -DKEEP_IT_REAL it would be a segoff_t equivalent.

Use this concept to avoid the need for bounce buffers in int13.c,
which reduces memory usage and opens up the possibility of using
multi-sector reads.

Extend the block-device API and the SCSI block device implementation
to support multi-sector reads.

Update iscsi.c to use user buffers.

Move the obsolete portions of realmode.h to old_realmode.h.

MS-DOS now boots an order of magnitude faster over iSCSI (~10 seconds
from power-up to C:> prompt in bochs).
2006-05-19 15:06:51 +00:00
Michael Brown 0ab92faedb Use typeof(sizeof(...)) to define a size_t. This stops gcc complaining
about format errors for %zX arguments.
2006-05-19 14:51:59 +00:00
Michael Brown 6b6096d28b Strip down i386 PCI configuration space I/O to the bare minimum. A
typical build will now include 880 bytes of PCI support code, compared to
2327 bytes in Etherboot 5.4.

(There is a slight cost of around 5 extra bytes per access to a
non-constant config space address; this should be an overall win.
Driver-specific accesses will usually be to constant addresses, for
which there is no additional cost.)
2006-05-17 01:12:11 +00:00
Michael Brown 15ee09ed10 Restructured PCI subsystem to fit the new device model.
Generic PCI code now handles 64-bit BARs correctly when setting
"membase"; drivers should need to call pci_bar_start() only if they want
to use BARs other than the first memory or I/O BAR.

Split rarely-used PCI functions out into pciextra.c.

Core PCI code is now 662 bytes (down from 1308 bytes in Etherboot 5.4).
284 bytes of this saving comes from the pci/pciextra split.

Cosmetic changes to lots of drivers (e.g. vendor_id->vendor in order to
match the names used in Linux).
2006-05-16 15:12:06 +00:00
Michael Brown fcdab6299c Missed a reference to heap.h. 2006-05-16 15:01:26 +00:00
Michael Brown 295e11b2b0 Add INT 13,41 (extensions installation check). LILO's MBR now uses
linear calls to load the MS-DOS boot sector in my test setup.
2006-05-14 23:46:28 +00:00
Michael Brown 29fb77b0a4 Fix a couple of minor typos. 2006-05-14 18:17:29 +00:00
Michael Brown 6e42cb95bd Also capture INT 19 as a failure path, to allow for boot sectors that
don't conform to BBS (e.g. the one created by mkdosfs).
2006-05-14 14:13:55 +00:00
Michael Brown 4435667a00 Add int13_boot(), to allow booting from INT 13 emulated drives. 2006-05-14 13:44:47 +00:00
Michael Brown b4f941b257 Specify attributes on .text16. 2006-05-14 13:39:27 +00:00
Michael Brown 2421723a15 Rewrote printf and friends to better support standard C semantics. 2006-05-14 02:51:55 +00:00
Michael Brown 5080668704 Provide access to rm_cs and rm_ds from protected-mode code. 2006-05-13 11:45:31 +00:00
Michael Brown dc8d429ec6 Provide a global counter for the number of interrupts hooked. 2006-05-13 11:44:53 +00:00
Michael Brown 63499e9cfd Allow re-entrancy and provide a global counter of the number of
interrupts hooked.
2006-05-13 11:44:11 +00:00
Michael Brown d822b3d4cd Define BIOS data segment in bios.h 2006-05-13 11:43:07 +00:00
Michael Brown a27defa483 Add flag definitions. 2006-05-13 11:42:07 +00:00
Michael Brown d2c22ec9bb Add INT 13 emulation layer (which provides an interface to gPXE block
devices).
2006-05-13 11:41:36 +00:00
Michael Brown 1ad72e0e79 Add 64-bit byte-swapping operations. 2006-05-13 11:40:39 +00:00
Michael Brown 1838b58c3f Add drivers/block and arch/i386/interface/pcbios 2006-05-13 11:35:29 +00:00
Michael Brown 568cad0ddc (Redoing check-in lost by SourceForge's failure.) 2006-05-13 11:30:03 +00:00
Michael Brown 847f38f4ac (Redoing check-in lost by SourceForge's failure.)
Add method for hooking real-mode interrupt vectors.
2006-05-13 11:19:06 +00:00
Michael Brown f7b963da51 (Redoing check-in lost by SourceForge's failure.)
Use .text16.data section with "aw" attributes, to avoid section type
conflicts when placing both code and data into .text16.

Add __from_{text16,data16}.
2006-05-13 11:11:55 +00:00
Michael Brown 97d265f8dc (Redoing check-in lost by SourceForge's failure.)
Avoid optimisation errors under -fstrict-aliasing.
2006-05-13 11:10:24 +00:00
Michael Brown fdb983d473 Preserve the whole of %esp across prot_call(). We have to split this
between the low half stored in the static variable rm_sp, and the high
half stored on the prot_call() stack, because:

  Just using the stack would screw up when a prot_call()ed routine
  executes a real_call(); it would have no way to find the current top of
  the RM stack.

  Extending rm_sp to rm_esp would not be safe, because the guarantee that
  rm_sp must return to the correct value by the time an external
  real-mode call returns applies only to %sp, not to %esp.
2006-05-06 19:44:23 +00:00
Michael Brown f8e087767b Allow access to variables in .text16 as well as .data16. Chained
interrupt vectors, for example, will be easiest to handle if placed in
.text16.
2006-05-06 18:57:31 +00:00
Michael Brown c10d1aa9d7 Preserve GDT across prot_call(). 2006-05-06 18:52:27 +00:00
Michael Brown dcc1b0054e This line should not have been checked in 2006-05-04 23:22:42 +00:00
Michael Brown 832807726a Add infrastructure to support access to .data16 (and .text16) variables
from protected-mode code.

Set up %ds to point to .data16 in prot_to_real, so that code specified
via REAL_EXEC() and friends can access variables in .data16.

Move most real-mode librm variables from .text16 to .data16.
2006-05-04 23:14:06 +00:00
Michael Brown f4429533a6 Added methods for efficiently declaring and accessing variables in
.data16.  librm will need to supply "char *data16", i.e. the virtual
address of the start of .data16.
2006-05-04 17:00:20 +00:00
Michael Brown f8f75cef5b BSS is now zeroed by libprefix (along with the otherwise non-zeroable
portions such as the stack) during the installation process.

init_fns() are called from main().  We need to think about initial
control flow some more.
2006-05-02 21:10:45 +00:00
Michael Brown cc23057b59 Typo 2006-05-02 21:09:29 +00:00
Michael Brown 10c28a51bd Create two easy-to-use entry points in libprefix: install and
install_prealloc.  I *think* these will suffice for all the 16-bit
prefixes.
2006-05-02 20:51:07 +00:00
Michael Brown 9e1dd6402d <name>_size variable is no longer needed, since we don't copy the code.
However, it's worth adding a .size directive so that objdump reports the
correct sizes for the code fragments.
2006-05-02 16:38:46 +00:00
Michael Brown 9fcded3d23 Towards a(nother) new real-mode infrastructure, in which we take
advantage of the fact that we have to have a permanently-resident block
in base memory.
2006-05-02 15:41:21 +00:00
Michael Brown 8f62b39c03 Move stack out to separate object, so that having a stack doesn't drag in
setup.S.
2006-05-02 14:57:48 +00:00
Michael Brown 5463169c1a Verified as working 2006-05-02 14:04:21 +00:00
Michael Brown d081d65d48 This should be much more elegant: we use flat real mode for the
highmem data, so decompress16 will be able to unpack blocks bigger
than 64kB.
2006-05-01 22:35:19 +00:00
Michael Brown cc8821a443 Checking in because I don't want to lose this rather neat code for
running the decompresser in 16:16 protected mode using the real-mode
stack.  However, there's an even simpler way to do it...
2006-05-01 21:26:44 +00:00
Michael Brown 9c9208a132 Put the TCP connection periodic processing in tcp.c, where it belongs. 2006-04-30 02:08:42 +00:00
Michael Brown 352bf1bda2 Move init.h to gpxe/init.h. 2006-04-30 01:08:52 +00:00
Michael Brown d8e99bf28f Gave up on adding POSIX errno's as required, and just added (almost) all
of them in one go.

EBADIMG has been replaced by ENOEXEC, and EIMGRET by ECANCELED.
2006-04-28 13:44:34 +00:00
Michael Brown 7a82c3eca3 Added missing ssize_t 2006-04-25 12:55:23 +00:00
Michael Brown 832e86246b gcc is rather over-aggressive about optimising out static data structures
even when __atribute__ (( unused )) is correctly set...
2006-04-24 18:31:37 +00:00
Michael Brown 4c4e4de18f (librm_base-1b) is already an offset; no need to apply OFFSET().
Doing so, in fact, seems to expose an assembler bug; (a-b-0) is
apparently not the same as (a-b).  Go figure.
2006-04-24 18:29:47 +00:00
Michael Brown 824d6ffa7f Header rearrangement.
I want to get to the point where any header in include/ reflects a
standard user-level header (e.g. a POSIX header), while everything that's
specific to gPXE lives in include/gpxe/.  Headers that reflect a Linux
header (e.g. if_ether.h) should also be in include/gpxe/, with the same
name as the Linux header and, preferably, the same names used for the
definitions.
2006-04-24 15:42:49 +00:00
Michael Brown 510ccb2900 Added __constant_memcpy() based on version in Linux's string.h.
Added memswap().
2006-04-24 15:25:25 +00:00
Michael Brown cdf1511d3e Prefix semantics have changed 2006-03-16 19:28:38 +00:00
Michael Brown 7e0a7a2e08 Merge from Etherboot 5.4 2006-03-16 18:30:54 +00:00
Michael Brown 9b18017296 Merge from Etherboot 5.4 2006-03-16 18:15:48 +00:00
Michael Brown bb7394b041 Merge from Etherboot 5.4 2006-03-16 17:59:51 +00:00