Gerd Zellweger [Thu, 4 Jul 2013 09:53:57 +0000]
FDIF functioning as a user level driver.
Gerd Zellweger [Wed, 3 Jul 2013 06:26:24 +0000]
Memory management functionality in driver.
Gerd Zellweger [Tue, 2 Jul 2013 13:57:09 +0000]
Towards user-level sdcard driver. Still need to add interrupts to I2C driver and enable register mapping in user-level.
Gerd Zellweger [Tue, 2 Jul 2013 07:24:49 +0000]
Getting PBIAS Interrupt consistently with the following steps:
a) Unplug Power cord and SDCARD
b) Plug in power
c) Plug in SD Card
d) Boot
e) Unplug SD Card
Gerd Zellweger [Tue, 2 Jul 2013 06:12:30 +0000]
Added debouncing register to the TWL 6030 mackerel file.
Gerd Zellweger [Fri, 28 Jun 2013 12:06:43 +0000]
Cleaning up debug print function.
Gerd Zellweger [Fri, 28 Jun 2013 12:05:37 +0000]
Fixed an issue in i2c API that does not ACK register read ready event.
Also added some debug prints.
Simon Gerber [Thu, 27 Jun 2013 09:17:55 +0000]
Changes to twl6030 Mackerel file to allow send register manipulations over i2c bus. TWL6030 driver to allow for voltage scaling.
Gerd Zellweger [Tue, 25 Jun 2013 18:11:49 +0000]
MMCHS driver cleanup
Using generated mackerel files in CM2.
Gerd Zellweger [Tue, 25 Jun 2013 13:23:59 +0000]
Some cleanup in gic_enable_interrupt.
Simon Gerber [Mon, 8 Jul 2013 09:30:04 +0000]
MMCHS and I2C code imported into main tree right now as a kernel-space library.
Needs to be changed to a user space driver.
Gerd Zellweger [Mon, 24 Jun 2013 07:47:34 +0000]
Added implementation for oct_del() in skb_simple on ARM.
USB implementation needs to delete records in case devices are detached.
Gerd Zellweger [Tue, 25 Jun 2013 07:51:35 +0000]
Make GCC binary for pandaboard_image depend on ARM_GCC variable in Makefile.
Gerd Zellweger [Mon, 10 Jun 2013 11:51:45 +0000]
(library) driver for face detection that can run inside the kernel right now.
Gerd Zellweger [Mon, 10 Jun 2013 07:18:49 +0000]
Fixed binary selection in build_data_files.sh.
Merge this with build fixes commit.
Gerd Zellweger [Wed, 5 Jun 2013 13:13:52 +0000]
Complete set of Mackerel Files for the OMAP4460 platform added.
These files have been auto-generated based on XML files found in
the TI RDT v1.0.0.4p tool (download at http://www.ti.com/product/omap4460).
The files will most likely compile. However, because their XML format
is less expressive than Mackerel some renaming might be helpful before use.
Especially to save some typing (i.e., remove duplicated namings, constant names etc.).
Also, some constants may be duplicated if their descriptions was
slightly different.
Gerd Zellweger [Thu, 6 Jun 2013 15:12:44 +0000]
Build fixes for ARMv7.
Make choice about GCC binary. Fixed typos in usbconf mackerel file.
Remove MMU from kernel Hakefile because it is not needed.
Don't do stack protection (leads to compiler errors on newer versions).
Don't add build-id when linking on ARMv7. This screwed up ELF loading in our
kernel when it tried to load a section at 0x8000 -- which is not a valid
virtual address -- for the init binary. For some reason
stripping the section manually did not help.
Achermann Reto [Mon, 8 Jul 2013 09:39:26 +0000]
Passing IO Capability to the monitor.
Achermann Reto [Tue, 2 Jul 2013 11:35:52 +0000]
Added arm_irq_handle message to monitor interface.
Achermann Reto [Tue, 2 Jul 2013 11:29:32 +0000]
Allow driver start-up on ARM by relying on Kaluga.
Achermann Reto [Tue, 2 Jul 2013 11:21:03 +0000]
Enable user-level interrupt forwarding on ARM.
Simon Gerber [Mon, 1 Jul 2013 11:17:20 +0000]
Added a section on connecting the tracing framework to Aquarium 2 directly in the tracing tech-note.
Stefan Kaestle [Mon, 24 Jun 2013 13:12:03 +0000]
Added brie machines to scalebench
Stefan Kaestle [Mon, 24 Jun 2013 12:06:09 +0000]
Merge
Stefan Kaestle [Mon, 24 Jun 2013 11:56:20 +0000]
Added basic usage information for Tracing framework
Simon Gerber [Mon, 24 Jun 2013 11:43:29 +0000]
Changed default bfscope port to 6666 to make everyone's lifes easier.
Stefan Kaestle [Fri, 7 Jun 2013 05:02:51 +0000]
adding a tag for release 2013-06-07
Kornilios Kourtis [Thu, 6 Jun 2013 06:13:14 +0000]
merge
Kornilios Kourtis [Thu, 6 Jun 2013 06:12:33 +0000]
symbolic_targets.mk: mkdir first when generating romfs_size.h
There seems to be a race in parallel builds:
echo "size_t romfs_cpio_archive_size = 0; //should not see this" > armv5/include/romfs_size.h
/bin/sh: cannot create armv5/include/romfs_size.h: Directory nonexistent
Simon Peter [Wed, 5 Jun 2013 02:31:54 +0000]
Removed debug.gdb.in.
Simon Peter [Wed, 5 Jun 2013 02:31:37 +0000]
New debug scripts.
* Removed a lot of old, unused cruft
* This does not depend on a pre-processed debug.gdb.in anymore
* Added example debugsim.sh
* Uses user-specific TCP port for GDB->QEMU connection
Kornilios Kourtis [Tue, 4 Jun 2013 13:02:31 +0000]
e1000: add octopus flounder defs/bindings
Kornilios Kourtis [Mon, 3 Jun 2013 14:45:23 +0000]
merge
Stefan Kaestle [Mon, 3 Jun 2013 10:19:55 +0000]
Check if child processes exist before terminating them
Stefan Kaestle [Mon, 3 Jun 2013 10:18:35 +0000]
Added omap44xx_mmu again to fix dependency error when building arm_gem5_image.
Improved code formatting.
Stefan Kaestle [Fri, 31 May 2013 13:46:51 +0000]
Dummy commit
Wang Nan [Fri, 31 May 2013 12:39:44 +0000]
armv7: disable vfp instructions for the cpu driver
Some compiler generates vfp instructions such as vpush, vldr for normal code in
cpu driver, which introduces complexity (related to fpu state). I propose a
patch which disables vfp instructions in cpu driver code.
Wang Nan [Fri, 31 May 2013 12:11:44 +0000]
armv7: align exception stack with 8 bytes
Here I propose a small patch which adjust arm exception stacks.
You can try 'printf("0x%016llx", 0x12345678ULL)' in undef handler. ARM abi
require stack be 8 bytes aligned, but original code setups 4 bytes gaps
between 2 stacks, makes irq_stack and undef_stack misaligned.
reference: http://infocenter.arm.com/help/index.jsp?topic=/com.arm.doc.faqs/ka4127.html
Gerd Zellweger [Thu, 30 May 2013 09:35:03 +0000]
Auto append the current year to the results path for a test.
This should avoid the "OSError: [Errno 31] Too many links" error in future tests.
We ran into this because we hit the ext3 limit in maximum subdirectories (32000)
in our test result directory. For now, lets hope we don't have more than 32000
test runs per year.
Gerd Zellweger [Thu, 30 May 2013 09:34:39 +0000]
PEP-8 Autoformatting.
Kornilios Kourtis [Wed, 29 May 2013 08:25:52 +0000]
merge
Gerd Zellweger [Wed, 29 May 2013 07:48:22 +0000]
Using platform independent size specifier 'z' in printf() for size_t.
Makes sure we can compile for 32bit.
Gerd Zellweger [Fri, 24 May 2013 13:38:54 +0000]
Don't write txce register in PHY mode.
Don't write asde register for 8257X variants.
Gerd Zellweger [Fri, 24 May 2013 08:51:20 +0000]
Use 82575 auto-negotiation for 82576 cards as well -- even though it already worked for these cards with the old code.
Gerd Zellweger [Fri, 24 May 2013 08:32:57 +0000]
Link-up routine for 82575 cards.
Gerd Zellweger [Fri, 24 May 2013 06:51:35 +0000]
Set bsize to 0 for 82575 cards (i.e., use 2048 bytes receive buffers).
This avoids splitted packets in the receive handler.
Gerd Zellweger [Wed, 22 May 2013 14:44:21 +0000]
Write Interrupt interval in extended register for 82575 cards.
Gerd Zellweger [Fri, 24 May 2013 06:49:21 +0000]
bsex bit does no longer exist in 82575 cards.
Gerd Zellweger [Fri, 24 May 2013 06:44:29 +0000]
Use different rxdctl register for 82575 cards.
Gerd Zellweger [Fri, 24 May 2013 07:17:32 +0000]
Set e1000_glci as default case for all cards that are not special cases.
Gerd Zellweger [Fri, 24 May 2013 06:39:06 +0000]
Don't write int_tca register for 82575 variants.
Gerd Zellweger [Fri, 24 May 2013 06:37:25 +0000]
Set IPG Register for 82575 card variants.
Gerd Zellweger [Thu, 23 May 2013 16:09:45 +0000]
Added another version for the txdctl register for the 82575 cards.
This is especially important because the new cards have the enable bit inside this register to enable/disable queues.
The code for the older cards set this bit to 0, therefore disabling the transmit queue.
Gerd Zellweger [Wed, 22 May 2013 11:03:36 +0000]
Fixed false register name in e1000 mackerel file (was bst, now cold) and adjusted the function call in the C code.
Simon Peter [Thu, 16 May 2013 06:51:12 +0000]
Fix Flounder messaging (TX_BUSY) for iref allocation inside the monitor.
Simon Gerber [Tue, 21 May 2013 14:11:10 +0000]
Moved mmap region resizing to general mmap formatting code.
Stefan Kaestle [Thu, 16 May 2013 13:51:50 +0000]
Make IO APIC work on machines where its memory map entry is smaller than a page.
On some machines (brie1) the IOAPIC region is only 1kB. Currently we are not able to map regions that are less than the page size so we increase the size of the region for such cases to one page.
Gerd Zellweger [Tue, 14 May 2013 15:48:18 +0000]
Added licensing and formatted according to Barrelfish coding guidelines.
Gerd Zellweger [Tue, 14 May 2013 14:11:39 +0000]
Added e1000 support for the follwing e1000 cards:
- 82574L
- 82575EB
- 82576EG
The 82575EB, 82576EG are somewhat different and do not transmit
anything at the moment. However since they are in our rack we
might want to work on those at some point.
Fixed some issues encountered on our test machines:
- In case auto-negotiation fails (in our machines with 82572 cards),
unset bit for auto-negotiation and force the link manually.
- 82574L does not seem to have ee_pres bit set
However, if we don't read from the EEPROM we cannot
get a mac address for our card, therefore we force the
read for this device right now.
Gerd Zellweger [Tue, 14 May 2013 12:53:44 +0000]
Use Octopus trigger to watch for e1000 cards that might appear after e1000 has started.
This makes sure e1000 works without using Kaluga. With the old code a race between PCI
and e1000 was possible (ie. e1000 looks for device records before PCI has added them).
Mateusz Olczak [Tue, 14 May 2013 15:46:06 +0000]
I've finished the work on the e1000 driver. All previous out standing issues have now been resolved.
* e1000 driver now works under Qemu, Simics and VirtualBox.
* Extended the number of supported NICs.
* Moved the e1000 driver to Mackerel 2 interface.
* e1000n_hwinit.c:
* e1000 driver now takes in consideration what card you have. Handles different cards more correctly.
* EEPROM: Extended the set of cards where eeprom reads are handled correctly for that card.
* Extended this set of supported cards.
* Checking that EEPROM read is done.
* Bailout and inform if EEPROM is not present.
* New read MAC address code.
* New card reset code.
* MAC address: Was not programmed correctly.
* Setting serial interface correctly for more cards. (previously same setting was made for all cards.)
* Link auto-negotiation. (previously link type and speed was forced disregarding if it was supported by media)
* Correctly setting hardware supported receive buffer size.
* Setting Transmit Inter Packet Gap (TIPG) correctly for more cards.
* Setting Interrupt Throttling to recommended values.
* e1000.dev: added register and updated various registers.
* e1000n.c: Removed uncommented code.
* Renamed some variables and function to make clear what they do.
* Added #ifdefine for parts that wore commented out but marked as "useful" for some purpose, such as CONFIG_MICROBENCHMARKS.
* Added probing for card type.
* Added help option.
* Added comments/documentation.
* Print of detected link speed and type.
* Updated printing of debug messages and ordinary messages.
* Fixed wrong max packet receive length for ethernet. (previous value was the max transmit size - They differ due to CRC)
* Other bug fixes minor.
Gerd Zellweger [Wed, 24 Apr 2013 14:05:19 +0000]
Make lhs2tex work again. It got confused with some commented code.
Simon Gerber [Mon, 13 May 2013 12:09:19 +0000]
Added quick-fix for ticket #253: Depending on total multiboot module size we run out of slots in the root cnode
Right now we just waste a couple kB of RAM to align the first RAM cap we
create to 1MB. A proper fix would involve increasing the size of the root
cnode.
Simon Gerber [Mon, 13 May 2013 09:25:54 +0000]
Cleanup trace.h.
Simon Gerber [Wed, 27 Mar 2013 14:20:09 +0000]
arm_gem5: cleaned up formatting in arm_gem5/init.c.
Gerd Zellweger [Tue, 7 May 2013 11:54:35 +0000]
Harness test creates a file called bootlog.txt containing only Barrelfish console output.
Useful because the BIOS output usually comes with some weird
characters that can mess up your editor.
Gerd Zellweger [Tue, 7 May 2013 11:39:02 +0000]
PEP-8 Autoformat for harness.py.
Gerd Zellweger [Sat, 13 Apr 2013 13:41:36 +0000]
Resolve namespace conflict in compute_required_resources.pl
ic solver and eclipse both provide a min/2 function which leads to
conflict when loading bridge_programming and compute_required_resource.
Wang Nan [Tue, 30 Apr 2013 12:14:30 +0000]
elver: disable SIMD instructions when building elver for x86_64
I discoverd that some gcc generages SIMD instructions for elver, which causes
problems, because when elver processing the kernel, SIMD feature is disabled.
Claudio Föllmi [Mon, 29 Apr 2013 11:30:58 +0000]
fixed single_slot_alloc_init_raw not compiling for arm
put in a check if we are compiling for arm, since __builtin_return_address does not support arguments !=0
Stefan Kaestle [Fri, 26 Apr 2013 12:46:00 +0000]
adding a tag for release 2013-04-26
Stefan Kaestle [Fri, 26 Apr 2013 12:43:39 +0000]
Fixed calculation of time offset between cores for Tracing Framework
Stefan Kaestle [Tue, 23 Apr 2013 12:12:01 +0000]
Added table of contents for Tracing Tech Note
Stefan Kaestle [Tue, 23 Apr 2013 09:09:11 +0000]
Fixed typesetting of an include command in the Tracing TechNote
Kornilios Kourtis [Tue, 2 Apr 2013 12:40:51 +0000]
fix buflen when calling single_slot_alloc_raw()
Here's my understanding (this is from reading the code):
- single_slot_alloc manages slots for a single cnode using a list of
free nodes. The list is maintained starting with ->head, and each node
represents a range of slots (slot, space). The list is initialized
with a single node with {slot=0, space=nslots}
- The worst case scenario for the size of the list is when only the odd
(or even) slots are free. This is the worst case because for all nodes
in the list space=1. Adding nodes should result in smaller number of
nodes but with higher ->space values.
- The list nodes are allocated/freed using a (simple) slab allocator.
Hence, the slab allocator should be initialized so that it contains
nslots/2 objects to accommodate for the worst case.
The problem is that this does not happen: the slab allocator requires some
additional memory for adding headers to objects which is not accounted for.
This is why in your case the allocator does not have additional objects to
provide in slab_alloc().
This commit allocates appropriately sized buffers for the slab allocator and
hopefully fixes your problem.
Many thanks to Georgios Varisteas <yorgos@kth.se> for reporting the bug and
help us debug the problem.
Kornilios Kourtis [Wed, 10 Apr 2013 11:27:41 +0000]
slab: do not use extra space for block header
Gerd Zellweger [Mon, 8 Apr 2013 14:34:21 +0000]
Do proper filtering of unneeded files in Hake. This avoids running into stack overflow: use +RTS -K<size> to increase it when having a build directory inside your tree.
Gerd Zellweger [Mon, 8 Apr 2013 11:42:44 +0000]
Disable kvm using the -no-kvm option of QEMU. Avoids having to unload / blacklist the module all the time in your Linux distribution.
Wang Nan [Tue, 2 Apr 2013 13:02:30 +0000]
hake: make hake work with ghc 7.4.2
Wang Nan [Tue, 2 Apr 2013 12:43:05 +0000]
tracing: fix trace_defs.h dependency
Kornilios Kourtis [Tue, 26 Mar 2013 19:11:56 +0000]
kaluga/octopus: make it compile for arm
Kornilios Kourtis [Tue, 26 Mar 2013 16:37:44 +0000]
thc: port eager THC to ARMv7
- I assume ARMv7, haven't look at possible differences with ARMv5
- thctest, thcminitest, thcidctest seem to work on pandaboard
- lazy THC is more tricky, I've added some notes for now
- should be possible to improve/optimize the functions
Kornilios Kourtis [Tue, 26 Mar 2013 16:23:50 +0000]
thc/x86: comments/cleanups to help ARM porting
Kornilios Kourtis [Tue, 26 Mar 2013 16:54:13 +0000]
add a spawn_wait_coreid() function
Patch by Mateusz Olczak <mateusz@olczak.se>
Kornilios Kourtis [Tue, 26 Mar 2013 16:47:48 +0000]
lshw: minor fix
Patch by Mateusz Olczak <mateusz@olczak.se>
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Fixed syntax error in omap44xx file.
---
devices/omap/omap44xx_mmu.dev | 14 +++++++-------
1 file changed, 7 insertions(+), 7 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Fix indenting.
---
devices/omap/omap44xx_mmu.dev | 356 ++++++++++++++++++++---------------------
1 file changed, 178 insertions(+), 178 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Register names conform with manual, completed bit description for all registers.
---
devices/omap/omap44xx_mmu.dev | 36 +++++++++++++++++++++++-------------
1 file changed, 23 insertions(+), 13 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
More work on MMU mackerel file.
---
devices/omap/omap44xx_mmu.dev | 81 ++++++++++++++++++++++++++++++++++-------
1 file changed, 68 insertions(+), 13 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
More complete MMU file.
---
devices/omap/omap44xx_mmu.dev | 78 +++++++++++++++++++++++++++++++++++------
1 file changed, 67 insertions(+), 11 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Fixed syntax error in mackerel file.
---
devices/omap/omap44xx_mmu.dev | 25 +++++++++++++------------
1 file changed, 13 insertions(+), 12 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Some more changes to omap44xx MMU mackerel files.
---
devices/omap/omap44xx_mmu.dev | 24 +++++++++++++++++++-----
1 file changed, 19 insertions(+), 5 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Add pandammu file to kernel Hakefile.
---
kernel/Hakefile | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Add pandammu.
---
devices/Hakefile | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Fix whitespace in devices Hakefile.
---
devices/Hakefile | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
Gerd Zellweger [Thu, 4 Apr 2013 12:27:12 +0000]
Initial Mackerel File for the Pandaboard MMU.
Contains a complete list of all the registers, however it is not complete in terms
of the individual register mappins.
---
devices/omap/omap44xx_mmu.dev | 81 +++++++++++++++++++++++++++++++++++++++++
1 file changed, 81 insertions(+)
create mode 100644 devices/omap/omap44xx_mmu.dev
Stefan Kaestle [Fri, 22 Mar 2013 18:53:00 +0000]
TN-008 not building with blessed toolchain when bitwidth argument given to bytefield environment
Stefan Kaestle [Fri, 22 Mar 2013 12:15:31 +0000]
adding a tag for release 2013-03-22
Kornilios Kourtis [Wed, 20 Mar 2013 17:29:06 +0000]
lshw/Hakefile: add flounderTHCStubs
Kornilios Kourtis [Wed, 20 Mar 2013 16:08:07 +0000]
trace: make it compile for x86_32