david/ipxe
david/ipxe
Archived
1
0
Fork 0
Code
2,545 Commits 1 Branch 0 Tags 14 MiB
6f39dfe6c0
Commit Graph

2493 Commits

Author SHA1 Message Date
Alexey Zaytsev
1935439f86 fix the rdtsc namimg 2008-03-02 05:19:29 +03:00
Alexey Zaytsev
379c37bafb Cleanups 
Replace a printf with a DBG in timer_rtdsc.c
Replace a printf in timer.c with assert
Return proper error codes from timer drivers

Signed-off-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
 2008-03-02 04:36:50 +03:00
Alexey Zaytsev
a1572e0ab0 Modify gPXE core and drivers to work with the new timer subsystem 
Signed-off-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
 2008-03-02 03:41:10 +03:00
Alexey Zaytsev
4006d229e5 Introduce the new timer subsystem. 
Timer subsystem initialization code in core/timer.c

	Split the BIOS and RTDSC timer drivers from i386_timer.c

	Split arch/i386/firmware/pcbios/bios.c into the RTSDC
	timer driver and arch/i386/core/nap.c

	Split the headers properly:
		include/unistd.h - delay functions to be used by the
					gPXE core and drivers.

		include/gpxe/timer.h - the fimer subsystem interface
					to be used by the timer drivers
					and currticks() to be used by
					the code gPXE subsystems.

		include/latch.h	- removed
		include/timer.h - scheduled for removal. Some driver
					are using currticks, which is
					only for core subsystems.

Signed-off-by: Alexey Zaytsev <alexey.zaytsev@gmail.com>
 2008-03-02 03:15:07 +03:00
Michael Brown
844828cb15 [DHCP] Fix RFC4390 client identifier constructions. 
RFC 4390 provides for the DHCP client identifier to contain the link-layer
hardware type and MAC address when the MAC address exceeds 16 bytes.
However, the hardware type field is only 8 bits; we were assuming 16 bits.
 2008-02-27 16:55:46 +00:00
Michael Brown
bc3f0a25c6 [Infiniband] Add preliminary support for multi-port devices. 
Arbel and Hermon cards both have multiple ports.  Add the
infrastructure required to register each port as a separate IB
device.  Don't yet register more than one port, since registration
will currently fail unless a valid link is detected.

Use ib_*_{set,get}_{drv,owner}data wrappers to access driver- and
owner-private data on Infiniband structures.
 2008-02-27 13:59:56 +00:00
Michael Brown
6570203571 [Infiniband] Centralise MAD operations 
Pull out common code for handling management datagrams from arbel.c
and hermon.c into infiniband.c.

Add port number to struct ib_device.

Add open(), close() and mad() methods to struct ib_device_operations.
 2008-02-26 22:56:19 +00:00
Michael Brown
03d1fc1da3 [Hermon] Donate joint copyright on Hermon driver to Mellanox. 2008-02-23 11:05:46 +00:00
Michael Brown
396dc6e797 [Hermon] Add driver for Mellanox Hermon (ConnectX) cards. 2008-02-23 10:51:38 +00:00
Michael Brown
ea5ad4be91 [Infiniband] Fix typo in debug statement 2008-02-23 10:48:18 +00:00
Michael Brown
563955ce37 [Infiniband] Add missing list_del()s in ib_create_qp() failure path. 2008-02-23 10:48:10 +00:00
Michael Brown
604c934981 Add DBGLVL_IO to trace all memory-mapped I/O. 2008-02-21 12:44:09 +00:00
Michael Brown
b87904ab20 Remove reference to COFF support. I can find no trace of ever having supported it. 2008-02-21 04:11:56 +00:00
Marty Connor
e6ff84b2d1 Make tarball generation quieter 
Deleting bin/deps after "make veryclean" keeps deps from being
regenerated in the tarball.
 2008-02-14 17:40:45 -05:00
Marty Connor
c9593cd35b Updated $(VERSION) for release. Change tarball generation. 
Added 'install' and 'configure' targets for those who are used
to other build setups.
 2008-02-14 17:26:47 -05:00
Michael Brown
c60050614a Added mtnic driver provided by Mellanox. 
Stripped out trailing whitespace to keep git happy.
 2008-02-13 15:01:18 +00:00
Michael Brown
385b7a623d Guard against corruption of top half of %esp during UNDI ISR 2008-02-13 14:58:20 +00:00
Michael Brown
c1006ffdb7 If no shell was requested via Ctrl-B, exit immediately if boot fails. 2008-02-11 17:51:44 +00:00
Michael Brown
40990465f1 Merge branch 'master' of rom.etherboot.org:/pub/scm/gpxe 2008-02-11 16:33:37 +00:00
H. Peter Anvin
fc9f016be0 UNDI ISR: save and restore 32-bit registers 
As written, if the if the UNDI ISR call clobbers the upper halves of
any of the GPRs (which by convention it is permitted to do, and by
paranoia should be expected to do) then nothing in the interrupt
handler will recover the state.

Additionally, save/restore %fs and %gs out of sheer paranoia - it's a
cheap enough operation, and may prevent problems due to poorly written
UNDI stacks.
 2008-02-11 15:09:47 +00:00
Michael Brown
f8a0b5f682 Use internal 16-bit stack for added safety. 2008-02-11 15:06:26 +00:00
H. Peter Anvin
9aec835541 undiisr.S: save/restore upper half of %eflags 
Since we don't know what the UNDI code does, it is safest to
save/restore %eflags even though the lower half of %eflags is
automatically saved by the interrupt itself.
 2008-02-10 18:13:39 -08:00
H. Peter Anvin
621c2886aa UNDI ISR: save and restore 32-bit registers 
As written, if the if the UNDI ISR call clobbers the upper halves of
any of the GPRs (which by convention it is permitted to do, and by
paranoia should be expected to do) then nothing in the interrupt
handler will recover the state.

Additionally, save/restore %fs and %gs out of sheer paranoia - it's a
cheap enough operation, and may prevent problems due to poorly written
UNDI stacks.
 2008-02-10 18:06:10 -08:00
Michael Brown
428c6342bc Fixes for EqualLogic iSCSI targets: 
Allow port numbers in iSCSI redirection.

  Wait for SCSI status, not just the final data-in (which may be followed
  by an explicit SCSI Response PDU if the S bit is not set).
 2008-02-07 22:46:26 +00:00
Michael Brown
5e4e267177 Start restructuring pxe_tftp.c to cope with multicast (MTFTP) receives. 2008-02-02 15:59:32 +00:00
Michael Brown
105ebec4f3 Merge commit 'laptop/master' 2008-01-22 18:57:01 +00:00
Michael Brown
122abb50af Merge branch 'master' of rom.etherboot.org:/pub/scm/gpxe 2008-01-22 18:53:01 +00:00
Michael Brown
c9c8cdeb8b Add pxe_set_cached_filename() so that pxe_tftp.c can also update the 
stored DHCP packets.
 2008-01-22 18:51:12 +00:00
Michael Brown
84f99f781e Use XFER_INIT() macro. 2008-01-22 18:50:24 +00:00
Michael Brown
a2b4252d5a Add XFER_INIT() macro. 2008-01-22 18:48:47 +00:00
Michael Brown
f6a8158eed Make seek information part of the xfer metadata, rather than an entirely 
separate xfer method.

Add missing .alloc_iob entries to several xfer_interface_operations
structures.
 2008-01-08 16:46:55 +00:00
Michael Brown
74fd544101 Added the embedded pxelinux payload patch from hpa. 2008-01-08 15:51:36 +00:00
Michael Brown
3311169707 When the DHCP file/sname fields are empty, don't allow them to override 
their equivalent DHCP-option-specified values.
 2008-01-07 13:50:32 +00:00
Michael Brown
de29e5a39c Use .SECONDARY instead of .PRECIOUS for bin/%.tmp targets. 2008-01-07 13:15:12 +00:00
Michael Brown
82bae302b1 Fix off-by-one error (discovered by Shao Miller). 2008-01-07 13:12:43 +00:00
Marty Connor
a9b8197f64 Unconditionally set utility program variables in Makefile to allow CROSS_COMPILE to work. 
stappers and xl0 pointed out that gnu make sets some variables, so ?=
is ineffective in some cases where we use it.. Cross-compilation
requires that some variables can be overridden in the
src/$(ARCH)/Config file, so include that file _after_ utility program
variables are set.
 2008-01-02 14:07:52 -05:00
Marty Connor
af56768e98 Retain some intermediate object files, suggested by Geert Stappers and Alexey Zaytsev. 
src/bin/%.tmp files can be useful for debugging, so let's keep them.
 2007-12-28 14:33:11 -05:00
Michael Brown
f3265b4bf8 Add preliminary support for MTFTP. 2007-12-26 18:51:20 +00:00
Marty Connor
9d4e4dbc32 3c90x driver fix (legacy) from Geert Staapers 
From: Geert Stappers <stappers@stappers.nl>
To: etherboot-developers@lists.sourceforge.net
Subject: [Etherboot-developers] 3c90x polling again [patch]
Date: Thu, 29 Nov 2007 09:22:36 +0100
User-Agent: Mutt/1.5.16 (2007-06-11)

Hello,

gPXE didn't work on 3COM 905C Tornado cards for me.
It did transmit the DHCP request, but it didn't see the DHCP offer.

Adding debug print statements allready solved the problem.
Attached is a patch that has a cleaner delay then print statements.
The core of it is
-    for(i=0;i<40000;i++);
+    mdelay(1);

There was no research if the change is about a longer delay
or about code NOT being optimized away. It works for me   :-)

Cheers
Geert Stappers
 2007-12-20 11:53:00 -05:00
Marty Connor
ff476c04e9 Fix e1000 receive tail pointer (RDT) handling 
e1000 should now work in VMware.
 2007-12-17 13:07:29 -05:00
Michael Brown
bea8787a03 Kill off the now-obsolete *_fill_nic() functions. 2007-12-15 19:40:55 +00:00
Michael Brown
7e8624a26f Merge branch 'master' of rom.etherboot.org:/pub/scm/gpxe 2007-12-15 19:37:44 +00:00
Michael Brown
5172d7b266 Autodetect whether or not we need to be using -fno-stack-protector. 2007-12-15 19:37:21 +00:00
Michael Brown
1d7974182e Overwrite the device descriptor's IRQ number with whatever the legacy 
driver's probe() routine fills in in nic->irqno.  This is so that
non-interrupt-capable legacy drivers which set nic->irqno=0 will end
up reporting IRQ#0 via PXENV_UNDI_GET_INFORMATION; this in turn means
that the calling PXE NBP will (should) hook the timer interrupt, and
everything will sort of work.
 2007-12-15 19:32:32 +00:00
Marty Connor
bd2607305c Correct e1000 interrupt routine to enable on 1 disable on 0. duh. 
The e1000_irq() routine should (per mcb30) do enable on non-zero,
disable on zero.  This is not consistent in all drivers, so I'll
wait to update it when doing a global sweep.
 2007-12-13 11:20:45 -05:00
Marty Connor
1a867bfb39 Remove *_fill_nic() calls, and directly set nic->ioaddr and nic->irqno . 
This needs to be done manually because if the irq() routine is
implemented then we want something like "nic->irqno = pci->irqno;",
else we do "nic->irqno = 0;" nic->ioaddr may also need to be set
carefully.

Also added local variables to end of many files, for emacs indentation
to match kernel style (tab does 8 space indent).
 2007-12-13 11:08:40 -05:00
Michael Brown
9747c1a486 Remove the obsolete pre-aBFT AoE boot table. 2007-12-12 14:32:02 +00:00
Michael Brown
3ab571006e Yet another attempt to autodetect an appropriate "echo -e". 2007-12-07 14:25:02 -06:00
Michael Brown
27b414da81 Ignore .echocheck internal housekeeping file. 2007-12-08 00:51:24 +00:00
Michael Brown
a8bddfe153 Another (hopefully more robust) attempt to find a usable substitute 
for "echo -e" on the host system.
 2007-12-08 00:50:57 +00:00
Michael Brown
2c7a9e803c Remove the (unused) option to override _prefix_link_addr, 
_textdata_link_addr, _load_addr and _max_align in the linker scripts.
A bug in some versions of ld causes segfaults if the DEFINED() macro
is used in a linker script *and* the -Map option to ld is present.

We don't currently need to override any of these values; if we need to
do so in future then the solution will probably be to always specify
the values on the ld command line, and have the linker script not
define them at all.
 2007-12-07 01:34:11 -06:00
Michael Brown
cf895a8c7c Avoid Makefile syntax that requires make >= 3.81. 2007-12-07 04:14:06 +00:00
Michael Brown
3b58fc0baf Try to fix echo-detection to work on all systems that provide any 
suitable "echo -e" substitute.
 2007-12-07 03:35:32 +00:00
Michael Brown
1949641d10 Fix compiler warnings that appear only on OpenBSD. 2007-12-06 14:16:46 -06:00
Michael Brown
798f7f6b8e Kill off some warnings-problematic debug statements in ndp.c; the code 
is non-functional anyway.
 2007-12-07 01:51:08 +00:00
Michael Brown
6b0e147e0d Don't complain when callers provide too-short buffers for 
PXENV_GET_CACHED_INFO.  NTLDR does this.
 2007-12-07 01:06:37 +00:00
Michael Brown
698b5bc887 Fix a minor logical error in posix_io.c 2007-12-07 00:52:48 +00:00
Michael Brown
df868476e7 Various warnings fixups for OpenBSD with gcc-3.3.5. 2007-12-07 00:11:43 +00:00
Michael Brown
64787bad29 Fix prototype mismatch. 2007-12-06 23:35:37 +00:00
Michael Brown
be6ada2b7b Add missing format qualifier 2007-12-06 21:38:23 +00:00
Marty Connor
82a3e80645 Add copyright, clean up comments a bit, remove unnecessary code. 
Cleanups for e1000.c.
 2007-12-03 01:17:41 -05:00
Michael Brown
1de705e30f Reallocate memory for bitmaps only when necessary. 2007-11-29 19:10:10 +00:00
Michael Brown
423e9d72f3 udp_open_promisc() calls udp_open_common() with peer==NULL. 2007-11-28 11:56:14 +00:00
Michael Brown
3b1efba864 Add RFC2090 TFTP multicast support. 2007-11-28 10:57:26 +00:00
Michael Brown
f770744ee0 Quick hack to be able to accept transfers from servers that don't 
supply options.
 2007-11-25 16:06:29 +00:00
Michael Brown
838ecba131 Provide individually cached constructed copies of DHCP packets via 
PXENV_GET_CACHED_INFO.  If we dont do this, Altiris' NBP screws up; it
relies on being able to grab pointers to each of the three packets and
then read them at will later.
 2007-11-22 04:43:11 +00:00
Michael Brown
3a2473f123 Do not fill in the BufferLimit field in struct 
s_PXENV_GET_CACHED_INFO, because this field doesn't exist in earlier
versions of the PXE spec, and some callers don't allocate space for it.
 2007-11-22 00:31:21 +00:00
Michael Brown
b3abf25e3c Allow DHCP server to instruct gPXE to ignore ProxyDHCP (which will 
also avoid waiting for ProxyDHCP offers).

Also reduce the ProxyDHCP timeout, because it's already irritating me.
 2007-11-21 22:10:04 +00:00
Michael Brown
0becbf5fba Add ProxyDHCP support. 2007-11-21 21:51:43 +00:00
Marty Connor
fa0bd77270 Added netdev_nullify to natsemi_remove() 2007-11-21 15:45:08 +00:00
Michael Brown
de329a36d0 Removed old e1000 driver. 2007-11-21 15:37:58 +00:00
Michael Brown
b53ee1eb16 Merge commit 'origin/mdc-working' 2007-11-21 15:35:08 +00:00
Michael Brown
fb809da2df Remove some assumptions about DHCP obtaining only a single options block. 2007-11-21 04:48:18 +00:00
Michael Brown
8d18338ae7 Add PXE-required DHCP options to DHCPDISCOVER and DHCPREQUEST packets. 2007-11-21 03:57:30 +00:00
Michael Brown
26f3a09ccf Add UUID to DHCP request as option 97 (if available). 2007-11-21 03:29:53 +00:00
Marty Connor
2d60a3963f Fixed receive instability by not enabling interrupts. duh. 
I have successfully booted DOS via iSCSI and PXELINUX via TFTP with this
version of the e1000 driver.
 2007-11-20 21:33:42 -05:00
Michael Brown
899f5b8ab2 Added definition of a UUID and uuid_ntoa() (for debugging), and 
implemented smbios_get_uuid().
 2007-11-21 02:27:07 +00:00
Michael Brown
68c438954d Update SMBIOS internal API to be easier to use, and not to require 
potentially exorbitant amounts of stack space.
 2007-11-21 01:58:27 +00:00
Marty Connor
645a752bc2 This version now transmits and receives. 
There may still be an issue with memory handling, since it seems to
die ungracefully when ARP packets come in after loading a kernel.
Something to debug.
 2007-11-20 16:23:12 -05:00
Marty Connor
8973caa1e6 Autonegotiation fixes. NIC now autonegotiates to whatever the switch supports. 2007-11-20 12:06:22 -05:00
Marty Connor
e41550b0ba A version of the new e1000 driver that transmits. 
autonegotiation still needs work.  receive is not working yet.
 2007-11-20 10:23:32 -05:00
Michael Brown
89eaab79c8 Replace "Etherboot" strings with "gPXE". 2007-11-20 02:00:54 +00:00
Michael Brown
6855c357a5 Merge branch 'master' of rom.etherboot.org:/pub/scm/gpxe 2007-11-19 20:14:49 +00:00
Michael Brown
a1e243d504 Add bin/pci_VVVV_DDDD target to allow building ROM images without 
needing to know the gPXE internal ROM name.
 2007-11-19 20:14:20 +00:00
Marty Connor
8ce93f0876 Fix typo. 2007-11-18 11:12:54 -05:00
Marty Connor
e21391d7b5 Add e1000 directory to build, since the driver compiles cleanly. 2007-11-18 10:57:20 -05:00
Marty Connor
8b0d86914b Rename e1000_main.c to e1000.c to so we can type 'make bin/e1000.dsk' instead of 'make bin/e1000_main.dsk' 2007-11-18 10:54:42 -05:00
Marty Connor
048a234dce Add errfile definition for new e1000.c 2007-11-18 10:45:20 -05:00
Marty Connor
0fea19c398 Introduce files for new e1000 driver 2007-11-18 10:31:00 -05:00
Marty Connor
2b8f77aca1 Add src/drivers/net/e1000 as home of new e1000 driver 2007-11-18 10:28:25 -05:00
Marty Connor
8564711c20 Move current e1000 driver files to e1000-old/ directory to avoid collisions 2007-11-18 09:58:28 -05:00
Marty Connor
8bd670301c Add ERRFILE information for e1000_main and e1000_hw, part of the new e1000 driver 
It is likely we will rename e1000_main to e1000 when the new e1000 driver is debugged.
 2007-11-18 09:51:21 -05:00
Marty Connor
9a4ed0f2d5 Remove unnecessary variable. 2007-11-18 03:11:02 -05:00
Michael Brown
c194b0c4bf Work around a bug in the OpenSolaris iSCSI target. 
We didn't specify values for MaxRecvDataSegmentLength and
MaxBurstLength (to save space, since we were happy with the
RFC-defined default values of 8kB and 256kB respectively).  However,
the OpenSolaris target (incorrectly) assumes default values of zero
for these parameters.

The upshot was that the OpenSolaris target would get stuck in an
endless loop trying to send us the first 512-byte sector, zero bytes
at a time, and would eventually run out of memory and core-dump.

Fixed by explicitly specifying the default values for these two
parameters.
 2007-11-05 13:29:05 +00:00
Michael Brown
755cb8379d Allow space for the kernel's real-mode .bss. Previously we weren't 
allowing any space for this, which makes it surprising that bzImage
loading ever worked.
 2007-11-04 04:27:38 +00:00
Michael Brown
c482fef058 Add missing attribute(packed) to two structures. 2007-10-29 21:22:00 +00:00
Michael Brown
82124ea688 Merge branch 'master' of rom.etherboot.org:/pub/scm/gpxe 2007-10-29 17:22:27 +00:00
Michael Brown
1620b3512c Merge branch '3leaf' 2007-10-29 17:21:58 +00:00
Michael Brown
be7d32396f Added one more PCI ID 2007-10-22 16:24:09 +01:00
Michael Brown
338d45a61e Change ROM names to lower case. 2007-10-11 00:43:21 +01:00
Michael Brown
35ee7e4479 Ensure that empty e820 regions are skipped even at the end of the 
memory map.  (We achieve this by setting CF on the last entry if it is
zero-length; this avoids the need to look ahead to see at each entry
if the *next* entry would be both the last entry and zero-length).

This fixes the "0kB base memory" error message upon starting Windows
2003 on a SunFire X2100.
 2007-10-10 20:13:34 +01:00
Michael Brown
e445c2c07c Ensure that empty e820 regions are skipped even at the end of the 
memory map.  (We achieve this by setting CF on the last entry if it is
zero-length; this avoids the need to look ahead to see at each entry
if the *next* entry would be both the last entry and zero-length).

This fixes the "0kB base memory" error message upon starting Windows
2003 on a SunFire X2100.
 2007-10-10 20:12:25 +01:00
Michael Brown
8069598a99 Add a missing PCI ID that is tested as working. 2007-10-10 02:07:18 +01:00
Michael Brown
9ac0377fbb Force a netdevice poll in net_tx() before attempting to transmit 
packet, to work around the problem whereby small TX rings get backed
up because we haven't yet poll()ed for TX completions.
 2007-10-04 14:57:33 +01:00
Michael Brown
2303c45d65 Force a netdevice poll in net_tx() before attempting to transmit 
packet, to work around the problem whereby small TX rings get backed
up because we haven't yet poll()ed for TX completions.
 2007-10-04 14:46:56 +01:00
Michael Brown
56550e400e Redefine bzimage_exec_context::mem_limit to be the highest permissible 
byte, rather than the number of permissible bytes (i.e. subtract one
from the value under the previous definition to get the value under
the new definition).

This avoids integer overflow on 64-bit kernels, where
bzhdr.initrd_addr_max may be 0xffffffffffffffff; under the old
behaviour we set mem_limit equal to initrd_addr_max+1, which meant it
ended up as zero.  Kernel loads would fail with ENOBUFS.
 2007-09-28 01:23:06 +01:00
Michael Brown
14fb6ba189 Quick and very dirty hack to get r8169 driver working again, prior to 
any rewrite.
 2007-09-26 17:58:51 +01:00
Michael Brown
ed5ff61790 Standardise DBG_PRINTF()->DBG(), and remove custom definitions for 
DBG_PRINTF() and assert().
 2007-09-26 17:38:36 +01:00
Michael Brown
881dd8e102 Switch rm_ss and rm_sp back to being words; it'll make it less 
confusing to read the code.
 2007-09-25 20:16:32 +01:00
Michael Brown
42d96bcb07 Don't use the "rep ss movsb" trick to copy the RM stack to the PM 
stack; it breaks vmxassist.
 2007-09-25 20:14:20 +01:00
Michael Brown
0ed066bc50 rm_ss, rm_sp, and pm_esp don't need to be accessed from anywhere 
outside of librm.S.
 2007-09-25 19:20:38 +01:00
Michael Brown
b8a8eea850 Minor refactoring to eliminate duplication. 2007-09-25 19:06:38 +01:00
Michael Brown
dec325fe43 Use full protected mode for access to high memory within prefix, to 
work around limitations in real-mode virtualisation support on Intel
VT-capable chips.
 2007-09-25 18:01:15 +01:00
Michael Brown
a7eee64506 Added more verbose memory-map debugging. 
Added redundant call to get_memmap() in int13_boot() immediately
before jumping to boot sector, to assist in debugging.
 2007-09-21 18:33:17 +01:00
Michael Brown
0979e674bb Added more verbose memory-map debugging. 
Added redundant call to get_memmap() in int13_boot() immediately
before jumping to boot sector, to assist in debugging.
 2007-09-21 18:27:51 +01:00
Michael Brown
44461b1854 Merge branch 'master' into 3leaf-delivery 2007-09-21 01:27:03 +01:00
Michael Brown
1e677b2024 Merge branch 'master' of rom.etherboot.org:/pub/scm/gpxe 2007-09-21 01:26:06 +01:00
Michael Brown
5793dbe96f Infinband requires longer link-layer addresses and headers than Ethernet. 2007-09-21 01:23:57 +01:00
Michael Brown
2ee76d0c93 Added defines for Infiniband files (which can't be committed to the 
main tree yet, but reserving the error numbers now avoids problems later).
 2007-09-21 01:19:04 +01:00
Michael Brown
1af346b531 Minor debug message improvement. 2007-09-21 01:17:46 +01:00
Michael Brown
af303db75d Use RFC4390 whenever hardware address exceeds 16 bytes; this allows us 
to construct DHCP packets suitable for Infiniband.
 2007-09-21 01:15:58 +01:00
Michael Brown
9c264faa0c Add const attribute to byte-swapping functions 2007-09-21 01:15:06 +01:00
Michael Brown
7e3527a658 Don't get stuck in an infinite loop on negative integers! 2007-09-21 01:14:44 +01:00
Michael Brown
db124b1255 Add fls() for non-constant values. 2007-09-21 01:14:38 +01:00
Michael Brown
f09173326c Moved iobuf.h assertions outside the static inline functions, so that 
the assert message's file and line number gives some clue as to the
real location of the problem.

Added similar assertions to list.h.
 2007-09-21 01:13:22 +01:00
Michael Brown
0f60150c44 Reverted driver/net/mlx_ipoib to clean master state 2007-09-18 07:07:04 +01:00
Michael Brown
30717896f5 IB can't afford many RX buffers, because the MTU is so large. Reduce 
the TCP window to compensate.
 2007-09-18 07:02:41 +01:00
Michael Brown
2f23881856 Added arbel.c 2007-09-18 06:37:22 +01:00
Michael Brown
95498fa038 Remove the last remaining visible hack. 2007-09-18 06:26:31 +01:00
Michael Brown
982e4dd101 Separated out to a clean new drivers/infiniband directory. 2007-09-18 06:06:55 +01:00
Michael Brown
a5ec029d24 First version that works with SELF_INIT! 2007-09-18 05:46:49 +01:00
Michael Brown
8423b0b109 Gets most of the way through initialisation. 2007-09-18 05:39:34 +01:00
Michael Brown
def5ae9127 Added sw2hw_mpt 2007-09-18 04:45:55 +01:00
Michael Brown
da014080f9 INIT_HCA block now matches. 2007-09-18 04:20:41 +01:00
Michael Brown
57bf8d9905 Seems to get through the ICM mapping process 2007-09-18 03:58:59 +01:00
Michael Brown
4ea6223f5a ICM allocation code in place; about to start test-and-compare runs. 2007-09-18 02:57:49 +01:00
Michael Brown
f20cd6f721 ICM allocation code *should* now be complete. 2007-09-18 02:05:25 +01:00
Michael Brown
af88f60228 Started with the ICM partitioning code. 2007-09-18 01:49:39 +01:00
Michael Brown
c818d47e25 Restructured bits of initialisation (not testable at present) 2007-09-17 22:25:46 +01:00
Michael Brown
dc3165b746 Starting the firmware directly now works. 2007-09-17 21:59:41 +01:00
Michael Brown
0f62bcaa77 Start migrating the remaining initialisation steps. 2007-09-17 21:14:09 +01:00
Michael Brown
f3fcb53faf Don't use the mailboxless version of the HW2SW_CQ command; it seems to 
crash the machine.
 2007-09-17 19:47:19 +01:00
Michael Brown
12b4cb964b Removed more hacks, and adjusted number of queue entries for likely usage. 2007-09-17 19:17:55 +01:00
Michael Brown
11541b1e01 Remove hacks, and fix leaving the multicast GID. 2007-09-17 19:12:06 +01:00
Michael Brown
267a4483ab Added an almost obscene amount of debugging and assertion code while 
tracking down a bug that turned out to be a free_iob() used where I
needed a netdev_tx_complete().  This left the freed I/O buffer on the
net device's TX list, with bad, bad consequences later.

Also fixed the bug in question.
 2007-09-17 19:01:10 +01:00
Michael Brown
387a1a8556 Minor debug message improvement. 2007-09-17 18:57:32 +01:00
Michael Brown
2c56ede6f8 Moved iobuf.h assertions outside the static inline functions, so that 
the assert message's file and line number gives some clue as to the
real location of the problem.

Added similar assertions to list.h.
 2007-09-17 18:38:04 +01:00
Michael Brown
bdac591726 Some interesting packet corruption happening now. 2007-09-17 13:31:13 +01:00
Michael Brown
2ed1acb9e9 Broadcast GID is now calculated by IPoIB layer. 2007-09-17 11:50:25 +01:00
Michael Brown
32a7bbb1e3 Cleaned up some debug messages. 2007-09-17 10:56:21 +01:00
Michael Brown
ab191e45db Unicasts seem to be working. :) 2007-09-17 10:39:30 +01:00
Michael Brown
2209090120 Dead code disabling 2007-09-17 08:41:06 +01:00
Michael Brown
b3d3814c17 Obtains a response to the get path record! 2007-09-17 08:36:56 +01:00
Michael Brown
e05a8cd4de Use pkey table access to determine broadcast GID directly. 2007-09-17 06:35:21 +01:00
Michael Brown
f6f1f2b7bb Prepare for adding a metadata queue to IPoIB 2007-09-17 06:12:33 +01:00
Michael Brown
440e7926fb Dead code removal 2007-09-17 05:12:47 +01:00
Michael Brown
4e78a53cf2 IPoIB code separated out to ipoib.c. 2007-09-17 05:04:58 +01:00
Michael Brown
67836430e6 Read port GID directly using MAD IFC. 2007-09-17 02:54:15 +01:00
Michael Brown
3c6a6bdc5d Multicast join now works. 2007-09-17 00:24:44 +01:00
Michael Brown
96d0c75c00 Now transmits packets on our own allocated IPoIB queue pair. :) 2007-09-16 22:53:57 +01:00
Michael Brown
4ddb6570f8 Almost working with own-queue allocation. 2007-09-16 22:47:45 +01:00
Michael Brown
7e85f0d296 create_qp() and destroy_qp() now written (but not tested). 2007-09-16 20:54:21 +01:00
Michael Brown
251cc84ed6 Started implementing create_qp() and destroy_qp(). 2007-09-16 19:03:24 +01:00
Michael Brown
83a6cc8c9b Don't get stuck in an infinite loop on negative integers! 2007-09-16 17:39:29 +01:00
Michael Brown
6d15a193aa Add fls() for non-constant values. 2007-09-16 17:36:40 +01:00
Michael Brown
b21d4ca21e Revert to dev_priv/owner_priv scheme, rather than container_of; it 
makes it easier to put the generic allocation code into infiniband.c
 2007-09-16 17:25:15 +01:00
Michael Brown
e238bb1e43 destroy_cq() now implemented (not tested). 2007-09-16 14:39:51 +01:00
Michael Brown
18edcf6634 create_cq() implemented (but not tested). 2007-09-16 14:18:17 +01:00
Michael Brown
725a574042 Match doorbell layout to expected usage. 2007-09-16 13:10:15 +01:00
Michael Brown
5a43293c38 Started to add code for CQ creation 2007-09-16 12:52:58 +01:00
Michael Brown
156b409ccc Rearrange data structures to maximise embedding (and hence minimise 
the number of separate allocations that need to be done).
 2007-09-16 04:02:20 +01:00
Michael Brown
791f992657 Command interface now reasonably friendly. 2007-09-16 03:13:25 +01:00
Michael Brown
bf9bd93856 First (working) draft of command interface. 2007-09-16 01:44:57 +01:00
Michael Brown
baa885ee8e Kill off more dead code. 2007-09-15 23:36:41 +01:00
Michael Brown
37fc40bc8c post_recv() now works, and we can pass data on the IPoIB queue pair 
using entirely our own code.
 2007-09-15 23:33:25 +01:00
Michael Brown
838b972cd3 Kill off some dead code 2007-09-15 21:26:14 +01:00
Michael Brown
31b82ea1dd Hack up IB structures at start-of-day, instead of on each use. 2007-09-15 21:23:34 +01:00
Michael Brown
8deef093d9 Direct polling of TX completion queue now works. 2007-09-15 20:58:29 +01:00
Michael Brown
687afdcdd9 Add const attribute to byte-swapping functions 2007-09-15 18:44:35 +01:00
Michael Brown
6a791649f0 Updated MLX_* accessor macros to use implicit type information. 2007-09-15 18:44:09 +01:00
Michael Brown
a3a91fedc1 Started added poll_cq() verb. 
Started reworking MLX_EXTRACT(), MLX_POPULATE() etc. to automatically
determine type information.
 2007-09-15 15:40:35 +01:00
Michael Brown
37a036bd48 Map the whole of physical memory 2007-09-15 03:22:04 +01:00
Michael Brown
970951666f arbel_post_send() has been observed to transmit a packet! 2007-09-15 02:53:05 +01:00
Michael Brown
8b27da9de1 Gets a response out of the hardware. (An error completion, to be precise.) 2007-09-15 01:35:07 +01:00
Michael Brown
21d4ab3ce2 Prefix arbel-specific functions etc. with arbel_ 2007-09-15 00:27:09 +01:00
Michael Brown
38a73b55c4 Now at least compiles 2007-09-15 00:19:38 +01:00
Michael Brown
9d08b7c692 Starting to introduce an Infiniband device abstraction 2007-09-14 20:29:44 +01:00
Michael Brown
75fbc96f75 Remove some dead code 2007-09-14 11:23:06 +01:00
Michael Brown
08e8dfd801 Now handling TX completions in our poll loop. 2007-09-14 11:10:25 +01:00
Michael Brown
e69863b5fb Proof-of-concept to manually parse completion event 2007-09-14 10:18:09 +01:00
Michael Brown
e9df4f691e Add EX_FLD_BE() 2007-09-14 10:17:29 +01:00
Michael Brown
da23e8d287 Start constructing a generic poll() routine. 2007-09-13 17:47:14 +01:00
Michael Brown
30a19c3f1c Can now both send and receive packets. LL header format not yet 
fixed; still using a quick hack-up just to be able to pass through
data.
 2007-09-13 14:43:12 +01:00
Michael Brown
99411e36da Added netdev_priv() to ease Linux driver compatibility. 2007-09-13 13:34:38 +01:00
Michael Brown
eecf71e4a0 Removed file checked in accidentally 2007-09-13 12:57:25 +01:00
Michael Brown
03c90e183f Dump received packet, including GRH. 2007-09-13 03:15:27 +01:00
Michael Brown
7e4e5af462 Use RFC4390 whenever hardware address exceeds 16 bytes; this allows us 
to construct DHCP packets suitable for Infiniband.
 2007-09-13 01:53:04 +01:00
Michael Brown
771455653b Bugfix: DHCP message type should be a one-byte option... (d'oh) 2007-09-13 01:16:23 +01:00
Michael Brown
9e32e8e513 Now sends IP packets correctly. 2007-09-13 01:14:56 +01:00
Michael Brown
c0d99245a9 Bugfix: DHCP message type should be a one-byte option... (d'oh) 2007-09-13 01:13:57 +01:00