544fa25928
Rename e{malloc,realloc,free} to u{malloc,realloc,free}, to more obviously reflect the fact that they allocate and deallocate user memory (i.e. things reached through a userptr_t).
Michael Brown2007-01-16 08:36:42 +00:00
ff8528ea9a
Create and use async_block() macro; it cuts down on the visual overhead of blocking on asynchronous operations, when that isn't an important aspect of the code.
Michael Brown2007-01-16 08:10:54 +00:00
f11da20f25
When a network device is specified to tcpip_tx() or it's children, treat it as a fallback network device rather than an explicitly forced network device.
Michael Brown2007-01-16 04:09:13 +00:00
143d14614d
Quickly hack in DNS resolution as a proof of concept
Michael Brown2007-01-15 17:32:52 +00:00
9af12d5fba
A working DNS resolver (not yet tied in to anything)
Michael Brown2007-01-15 17:31:35 +00:00
df0397f334
Update TFTP and FTP to take the same temporary URI scheme as HTTP
Michael Brown2007-01-15 09:58:26 +00:00
afa752f5fb
Protocol's get() method no longer takes ownership of the URI. HTTP is the exception rather than the rule; we may as well keep things clean for other protocols.
Michael Brown2007-01-15 09:18:06 +00:00
1f92c6b3e8
Must free http on the error path; nothing else will do it
Michael Brown2007-01-15 09:08:40 +00:00
c676591cd1
Unmaintained example code is probably worse than useless.
Michael Brown2007-01-15 09:06:12 +00:00
4e20d73bb5
Gave asynchronous operations approximate POSIX signal semantics. This will enable us to cascade async operations, which is necessary in order to properly support DNS. (For example, an HTTP request may have to redirect to a new location and will have to perform a new DNS lookup, so we can't just rely on doing the name lookup at the time of parsing the initial URL).
Michael Brown2007-01-15 08:49:10 +00:00
26166cf5e0
Added a URI parser that should be standards conformant. (It can certainly handle something as convoluted as
Michael Brown2007-01-14 23:48:31 +00:00
a5a330339b
Split DBG() statement containing two inet_ntoa() statements; they return a static buffer.
Michael Brown2007-01-14 19:58:41 +00:00
ce7289a766
Dump buffer and packet information in PXENV_UDP_READ as well
Michael Brown2007-01-14 18:22:37 +00:00
76aa9ad07d
Make TCP give up immediately when it receives -ENETUNREACH from tcpip_tx(). This avoids the irritating wait when you accidentally type "kernel pxelinux.0" before bringing up the network interface.
Michael Brown2007-01-14 16:47:03 +00:00
c953c1a1c3
Use -ENETUNREACH to mean "no reachable network device exists, don't bother retrying".
Michael Brown2007-01-14 16:22:10 +00:00
a5f6408d8e
We can now load an initrd as well as a kernel
Michael Brown2007-01-14 16:09:01 +00:00
b9f99858ab
Make the getmemsize() prototype available in memsizes.h, for code that doesn't want to go to the hassle of processing a full memory map.
Michael Brown2007-01-14 15:32:25 +00:00
c1fd6e8dc7
Once image type has been set, do not autoprobe further
Michael Brown2007-01-14 15:25:23 +00:00
aaed3d50a8
Some operating systems get a nasty shock if a region of the E820 map seems to start on a non-page boundary. Make life safer by rounding out our edited regions.
Michael Brown2007-01-13 23:38:33 +00:00
d0f4e9a54b
Add some trace messages for important events
Michael Brown2007-01-13 17:54:41 +00:00
526d314266
Advertise a larger MSS to improve TCP performance.
Michael Brown2007-01-13 17:36:17 +00:00
4b77061881
Pick up the return status code from the correct place now that we don't overwrite the parameter block until *after* the debug code.
Michael Brown2007-01-13 16:55:57 +00:00
5817c9f114
When an UNDI API call fails, print everything there is to know about it.
Michael Brown2007-01-13 16:53:55 +00:00
3b77c7aa1b
Hex dumps are now integrated into the DBG() framework.
Michael Brown2007-01-13 16:49:38 +00:00
6a3c76c8e0
Always send EOI. We can't feasibly share interrupts (since we have no clue what the "previous" interrupt handler will do, which could range from "just an iret" to "disable the interrupt"), and that means that we have to take responsibility for ACKing all interrupts. Joy.
Michael Brown2007-01-13 14:45:26 +00:00
86171c53f4
Damn Broadcom and their damned incorrect assumptions about x86 memory allocation.
Michael Brown2007-01-13 01:48:12 +00:00
63386f4455
Added ENOENT, since HTTP 404 translates to it
Michael Brown2007-01-12 19:14:01 +00:00
ca3db0bf11
Added isspace() and made strtoul() accept whitespace, as per POSIX.
Michael Brown2007-01-12 19:11:28 +00:00
83b7933f8a
Damn it; my lovely resilient scheme falls down when you have a protocol that switches from line-oriented to byte-oriented partway through, such as HTTP.
Michael Brown2007-01-12 18:09:14 +00:00
ad22cccc09
No need to maintain a received byte count; we always fill in sequential order so we can just use buffer->fill.
Michael Brown2007-01-12 17:16:46 +00:00
996b091b50
Added generic line-buffering code (a la stdio)
Michael Brown2007-01-12 17:08:37 +00:00
bcd6ca3291
Place multiboot tables in base memory; at least some OSes won't see it if we don't.
Michael Brown2007-01-12 16:25:05 +00:00
5c72d9c4e0
Print image physical location in registration message.
Michael Brown2007-01-12 16:20:51 +00:00
0f29e0e46e
Ensure multiboot modules are in ascending order of memory start address, to work around OS bugs.
Michael Brown2007-01-12 12:03:19 +00:00
e2c0055e23
Let ifmgmt.c take care of calling efree(), since it's the once which took out the contract to eventually call efree() when it called fetch().
Michael Brown2007-01-12 09:46:10 +00:00
475d6d1f7c
fetch() now knows nothing about struct image; it simply loads a file and returns the allocated buffer.
Michael Brown2007-01-12 08:02:27 +00:00
2876197306
Allow "imgexec" with no arguments to boot the file that was loaded with "kernel".
Michael Brown2007-01-12 07:32:46 +00:00
f135a37f30
Cleaner separation between imgXXX() functions and image_cmd.c
Michael Brown2007-01-12 07:22:20 +00:00