8 years agoschedule_rbed.c: move include of timer.h so that simulator compiles release2011-11-10
Kornilios Kourtis [Wed, 9 Nov 2011 10:38:16 +0000]
schedule_rbed.c: move include of timer.h so that simulator compiles

8 years agoadd support for one-shot timers
Kornilios Kourtis [Wed, 9 Nov 2011 10:15:49 +0000]
add support for one-shot timers

- Only for x86_64, and disabled by default using a config variable
- changes in existing code are guarded by #ifdefs

8 years agoAdd some scheduling related comments
Kornilios Kourtis [Wed, 9 Nov 2011 09:47:55 +0000]
Add some scheduling related comments

8 years agoReverted commit 34798966b84b, which was mistakenly applied twice.
Simon Peter [Tue, 8 Nov 2011 22:14:20 +0000]
Reverted commit 34798966b84b, which was mistakenly applied twice.

8 years agousr/startd: allow comments (#) and empty lines in bootscript
Kornilios Kourtis [Mon, 26 Sep 2011 18:32:21 +0000]
usr/startd: allow comments (#) and empty lines in bootscript

9 years agoc/fflush(): if write buffer is empty, just return
Kornilios Kourtis [Mon, 19 Sep 2011 13:37:38 +0000]
c/fflush(): if write buffer is empty, just return

9 years agobugfix in check_for_event() that prevented the channel being polled
Kornilios Kourtis [Fri, 16 Sep 2011 13:46:38 +0000]
bugfix in check_for_event() that prevented the channel being polled

9 years agoimplemented timeout for select. fixed the wedging problem for postgres.
Jana Giceva [Tue, 13 Sep 2011 14:51:44 +0000]
implemented timeout for select. fixed the wedging problem for postgres.

9 years agoFixed domain spanning under memory reclamation: Need to mark custom
Simon Peter [Wed, 24 Aug 2011 12:59:24 +0000]
Fixed domain spanning under memory reclamation: Need to mark custom
created caps as remote.

8 years agoFixed two memory reclamation problems:
Simon Peter [Sat, 5 Nov 2011 20:51:14 +0000]
Fixed two memory reclamation problems:

1. Need to update the consumed field, otherwise vspace library leaks memory.

2. Have every domain keep up to 1MB of free memory around as fast reserve,
   before returning to memory server. This circumvents a problem with
   asynchronous memory reclamation: If memory is allocated and freed in a
   tight loop, allocating memory happens faster than it will be freed
   (freeing is an asynchronous operation). This commit works on the
   assumption that allocating and then freeing more than 1MB in a tight loop
   is not something one does regularly. Also, keeping a little bit of memory
   as local reserve should speed up fine-grain memory management.

8 years agoadded cryptotest to ensure the library is built.
Adrian Schuepbach [Mon, 10 Oct 2011 13:40:22 +0000]
added cryptotest to ensure the library is built.

8 years agoadded OpenSSL
Adrian Schuepbach [Mon, 10 Oct 2011 13:30:22 +0000]
added OpenSSL

8 years agominor tweeks to OpenSSL
Adrian Schuepbach [Mon, 10 Oct 2011 12:42:35 +0000]
minor tweeks to OpenSSL

8 years agoadded waiting mechaanism for the datagatherer.
Adrian Schuepbach [Wed, 5 Oct 2011 13:37:07 +0000]
added waiting mechaanism for the datagatherer.

8 years agoadded test algorithm for the Barcelona 2010 workshop.
Adrian Schuepbach [Wed, 5 Oct 2011 13:07:33 +0000]
added test algorithm for the Barcelona 2010 workshop.

8 years agoadded datafiles and formats as used in the ASPLOS 2011 paper
Adrian Schuepbach [Wed, 5 Oct 2011 13:06:19 +0000]
added datafiles and formats as used in the ASPLOS 2011 paper

8 years agoadded original code with modulo operation to find the natural alignment.
Adrian Schuepbach [Wed, 5 Oct 2011 13:01:00 +0000]
added original code with modulo operation to find the natural alignment.

8 years agoadded OpenSSL
Adrian Schuepbach [Wed, 5 Oct 2011 12:59:35 +0000]
added OpenSSL

8 years agoadded a few test programs.
Adrian Schuepbach [Wed, 5 Oct 2011 12:58:21 +0000]
added a few test programs.

8 years agoadded strcasecmp.c
Adrian Schuepbach [Wed, 5 Oct 2011 12:57:25 +0000]
added strcasecmp.c

8 years agoremoved an unused variable.
Adrian Schuepbach [Tue, 4 Oct 2011 14:40:10 +0000]
removed an unused variable.

8 years agoupdated db_partitions such that it does not overload the NUMA regions.
Adrian Schuepbach [Tue, 4 Oct 2011 13:34:20 +0000]
updated db_partitions such that it does not overload the NUMA regions.

8 years agorenamed labelall, because otherwise it replaces the one from bridge_page and we defin...
Adrian Schuepbach [Mon, 3 Oct 2011 19:40:28 +0000]
renamed labelall, because otherwise it replaces the one from bridge_page and we definitely don't want thathg di|less

8 years agoadded global thead allocation code.
Adrian Schuepbach [Mon, 3 Oct 2011 19:32:54 +0000]
added global thead allocation code.

8 years agodb_partitions based on globa thread alloc.
Adrian Schuepbach [Mon, 3 Oct 2011 19:32:26 +0000]
db_partitions based on globa thread alloc.

8 years agoinitial DB partitions code.
Adrian Schuepbach [Mon, 3 Oct 2011 14:02:59 +0000]
initial DB partitions code.

8 years agoadded cache query function.
Adrian Schuepbach [Mon, 3 Oct 2011 14:02:28 +0000]
added cache query function.

9 years agoFixed a small bug with the multihop messaging. Other interconnect drivers written...
Stefan Kästle [Fri, 16 Sep 2011 12:29:19 +0000]
Fixed a small bug with the multihop messaging. Other interconnect drivers written down after the multihop code intialization should be loaded as well. The return statement shouln't be there

9 years agoLatent weirdness in the XScale build process (which could probably be
Mothy [Wed, 7 Sep 2011 13:35:52 +0000]
Latent weirdness in the XScale build process (which could probably be
rationalized), together with further tweaking of device files.

rename : devices/ => devices/

9 years agoFixed more inconsistent device names, this time for ARM.
Mothy [Wed, 7 Sep 2011 12:41:03 +0000]
Fixed more inconsistent device names, this time for ARM.

rename : devices/ => devices/

9 years agoFixed mixed up Parsec's 2 and 3.
Mothy [Wed, 7 Sep 2011 12:09:42 +0000]
Fixed mixed up Parsec's 2 and 3.

9 years agoAdded gratuitous uhci device, for qemu support in the future.
Mothy [Wed, 7 Sep 2011 10:04:40 +0000]
Added gratuitous uhci device, for qemu support in the future.

9 years agomerge
Mothy [Wed, 7 Sep 2011 09:38:55 +0000]

9 years agoRemoved spurious device include.
Mothy [Wed, 7 Sep 2011 09:30:28 +0000]
Removed spurious device include.

9 years agoCleanup: removed redundant ( has it all).
Mothy [Wed, 7 Sep 2011 08:49:06 +0000]
Cleanup: removed redundant ( has it all).
Moved rtc in-kernel driver to use Shift driver.
Made it clearer that the bitfield driver is deprecated for new devices.
Fixed minor bug in Shift driver.

9 years add missing space in bash test
Kornilios Kourtis [Wed, 7 Sep 2011 07:50:34 +0000] add missing space in bash test

9 years agoRemove the dependency from fish on memclient and ramclient by moving the
akhi [Tue, 6 Sep 2011 13:02:08 +0000]
Remove the dependency from fish on memclient and ramclient by moving the
required functionality into lib/barrelfish.

This ensures that there is no duplicate flounder code generation for the two
clients in fish.

9 years agoNow compiling all X86 code; renamed identifiers for consistency.
Mothy [Tue, 6 Sep 2011 08:23:03 +0000]
Now compiling all X86 code; renamed identifiers for consistency.

rename : devices/ => devices/

9 years agoRenamed many devices which were inconsistent with their .dev file names.
Mothy [Fri, 2 Sep 2011 14:50:08 +0000]
Renamed many devices which were inconsistent with their .dev file names.

rename : usr/pci/LPC_IOAPIC_ioapic_impl.h => usr/pci/lpc_ioapic_ioapic_impl.h
rename : usr/pci/LPC_IOAPIC_spaces.h => usr/pci/lpc_ioapic_spaces.h

9 years agofish also depends on the memserv binding. There's a new command to report on release2011-09-02
Simon Peter [Fri, 2 Sep 2011 08:50:53 +0000]
fish also depends on the memserv binding. There's a new command to report on
free memory.

9 years agoLPC timer still needed for vmkit.
Simon Peter [Wed, 31 Aug 2011 14:06:52 +0000]
LPC timer still needed for vmkit.

9 years agomerge
Mothy [Wed, 31 Aug 2011 13:53:35 +0000]

9 years agoSorry, another test push
Stefan Kästle [Tue, 30 Aug 2011 16:19:49 +0000]
Sorry, another test push

9 years agoAnother test push
Stefan Kästle [Tue, 30 Aug 2011 16:11:53 +0000]
Another test push

9 years agoTest push
Stefan Kästle [Tue, 30 Aug 2011 15:48:47 +0000]
Test push

9 years agomerge
Mothy [Tue, 30 Aug 2011 09:56:10 +0000]

9 years agoTracked down the two figures needed in the routing documentation.
akhi [Tue, 30 Aug 2011 09:54:09 +0000]
Tracked down the two figures needed in the routing documentation.

9 years agoCleaned up missing references, etc.
Mothy [Tue, 30 Aug 2011 08:49:52 +0000]
Cleaned up missing references, etc.

9 years agoTemporarily exclude missing figures from the routing document.
akhi [Tue, 30 Aug 2011 08:12:45 +0000]
Temporarily exclude missing figures from the routing document.
They will be added once we track down the figures.

9 years agoremove some lwip debug printing
Andrew Baumann [Mon, 29 Aug 2011 17:35:55 +0000]
remove some lwip debug printing

9 years agoFixed domain spanning under memory reclamation: Need to mark custom
Simon Peter [Wed, 24 Aug 2011 12:59:24 +0000]
Fixed domain spanning under memory reclamation: Need to mark custom
created caps as remote.

9 years agomerged
akhi [Wed, 24 Aug 2011 13:07:11 +0000]

9 years agoBug fixes by Alex on his multihop code and updates to the routing tech note.
akhi [Wed, 24 Aug 2011 13:06:29 +0000]
Bug fixes by Alex on his multihop code and updates to the routing tech note.

9 years agoAdded address-space measurements for the SCC
Stefan Kästle [Mon, 22 Aug 2011 09:23:02 +0000]
Added address-space measurements for the SCC

9 years agomerge
Mothy [Sun, 21 Aug 2011 09:56:25 +0000]

9 years agoUpdated documentation.
Mothy [Sun, 21 Aug 2011 09:56:18 +0000]
Updated documentation.

9 years agoApply the multihop patch by alexander grest based on his bachelor thesis
akhi [Tue, 16 Aug 2011 15:40:23 +0000]
Apply the multihop patch by alexander grest based on his bachelor thesis

9 years agoremoved few debug prints [Mon, 15 Aug 2011 20:11:55 +0000]
removed few debug prints

9 years agoFixed the command line argument issue with vmkit test [Mon, 15 Aug 2011 20:07:53 +0000]
Fixed the command line argument issue with vmkit test

9 years agoFixed the bug in the continuation management code [Mon, 15 Aug 2011 19:56:39 +0000]
Fixed the bug in the continuation management code

9 years agoFixed broken merge
Mothy [Fri, 12 Aug 2011 21:20:13 +0000]
Fixed broken merge

9 years agomerge
Mothy [Fri, 12 Aug 2011 20:42:44 +0000]

9 years agoAdded TypeName (oops)
Mothy [Fri, 12 Aug 2011 20:42:18 +0000]
Added TypeName (oops)

9 years agoMerge
Mothy [Fri, 12 Aug 2011 20:41:22 +0000]

9 years agoCheckpoint: removing __DP functionality to make imports work as expected.
Mothy [Fri, 12 Aug 2011 15:25:43 +0000]
Checkpoint: removing __DP functionality to make imports work as expected.

9 years agoMerge.
Mothy [Thu, 11 Aug 2011 14:51:50 +0000]

9 years agoImports almost working for Mackerel; with some new restrictions:
Mothy [Thu, 11 Aug 2011 14:50:09 +0000]
Imports almost working for Mackerel; with some new restrictions:

1) #define dev_PREFIX is unlikely to work again
2) Incorrect DP() macros, which will be fixed before release
3) You have to name each file exactly in accordance with each device.

9 years agoICC compilation fixes
Andrew Baumann [Wed, 10 Aug 2011 23:03:49 +0000]
ICC compilation fixes

9 years agoquieten another printf which can occur if a packet arrives during startup and lead...
Andrew Baumann [Wed, 10 Aug 2011 21:08:39 +0000]
quieten another printf which can occur if a packet arrives during startup and lead to an infinite loop of printing

as discussed, this needs to be fixed properly by not enabling receive until we're ready to receive packets

9 years agonetd: warn the user and exit cleanly if a newly-required command-line option isn...
Andrew Baumann [Wed, 10 Aug 2011 20:38:54 +0000]
netd: warn the user and exit cleanly if a newly-required command-line option isn't present

FWIW, in the absence of a service discovery mechanism, I think we
should be using generic names like eth0 eth1 etc. to name network
interfaces, rather than having to wire driver names into every
application's command-line arguments.

9 years agodon't print debug output for an expected error
Andrew Baumann [Wed, 10 Aug 2011 20:37:33 +0000]
don't print debug output for an expected error

9 years agomerged
Andrew Baumann [Wed, 10 Aug 2011 19:11:20 +0000]

9 years agopci: reduce minimum chunking granularity in physmem allocator to 1 (i.e. 2 caps at...
Andrew Baumann [Wed, 10 Aug 2011 19:02:18 +0000]
pci: reduce minimum chunking granularity in physmem allocator to 1 (i.e. 2 caps at each level)

This wastes memory, but is needed to ensure that we can always
allocate caps to BARs at the correct granularity. I needed it to get
the framebuffer cap working on an Nvidia card; hopefully it doesn't
break other hardware :(

9 years agovbe: extended interface to support cards with a framebuffer in the middle of a BAR
Andrew Baumann [Wed, 10 Aug 2011 18:53:47 +0000]
vbe: extended interface to support cards with a framebuffer in the middle of a BAR

also cleaned up types: errval_t rather than int!

9 years agocompilation fixup after merge
Andrew Baumann [Wed, 10 Aug 2011 16:31:13 +0000]
compilation fixup after merge

9 years agomerged, manually
Andrew Baumann [Wed, 10 Aug 2011 16:22:38 +0000]
merged, manually

9 years agoFixed the lines which were causing problems in detecting the IP [Wed, 10 Aug 2011 09:38:43 +0000]
Fixed the lines which were causing problems in detecting the IP
address by harness.

9 years agoFixed the line which was causing problem in detection of IP address [Wed, 10 Aug 2011 09:23:37 +0000]
Fixed the line which was causing problem in detection of IP address

9 years agoe1000: trivial debug/error checking changes
Andrew Baumann [Wed, 10 Aug 2011 00:12:36 +0000]
e1000: trivial debug/error checking changes

9 years agoe1000: improvements to MAC programming/handling code allowing the driver to work...
Andrew Baumann [Wed, 10 Aug 2011 00:12:00 +0000]
e1000: improvements to MAC programming/handling code allowing the driver to work on more cards

1. If a  valid MAC is already present in  the card's config registers,
   skip trying  to read it from  EEPROM (which is  different between many
   models of e1000)  and just use the value  that's already present. This
   was enough to get the driver working on an 82567.
2. Allow the user to specify the MAC address as a command-line parameter.

9 years agolpc_kbd: wait for buffers to drain rather than just asserting that they did
Andrew Baumann [Wed, 10 Aug 2011 00:06:06 +0000]
lpc_kbd: wait for buffers to drain rather than just asserting that they did

(happens on real HW)

9 years agomonitor: remove unused variable
Andrew Baumann [Wed, 10 Aug 2011 00:05:20 +0000]
monitor: remove unused variable

9 years agonetd: fix format strings, extra assertions to catch filter compilation failures
Andrew Baumann [Wed, 10 Aug 2011 00:05:08 +0000]
netd: fix format strings, extra assertions to catch filter compilation failures

9 years agonetd: sanely handle and report changes to the IP address after initially configured
Andrew Baumann [Wed, 10 Aug 2011 00:04:44 +0000]
netd: sanely handle and report changes to the IP address after initially configured

... this can happen if the DHCP server decides to change our IP, or refuse a renewal

9 years agonfs: correctly handle mount failures
Andrew Baumann [Wed, 10 Aug 2011 00:03:34 +0000]
nfs: correctly handle mount failures

The error reporting and handling in the NFS API is really crappy
anyway, but at least pass the NULL in the right place to prevent a
page fault on mount failures...

9 years agodon't (unconditionally) print a message when dropping a packet
Andrew Baumann [Wed, 10 Aug 2011 00:02:45 +0000]
don't (unconditionally) print a message when dropping a packet

The specific problem that led to this was a network with a high rate
of broadcast traffic. We hit this printf if a packet arrives in the
window between netd connecting to e1000 and finishing buffer
registration. If the rate of packet arrival was higher than the rate
of printfs (not unusual with our serial driver), we'd loop forever
telling the user that we were discarding packets without actually
making progress.

9 years agobfdmuxtools: construct filters with %u not %d format strings
Andrew Baumann [Wed, 10 Aug 2011 00:00:57 +0000]
bfdmuxtools: construct filters with %u not %d format strings

... I had a problem where filter compilation was failing on a negative
number. Since all integers are parsed with strtoul or strtoumax, I
believe the only sane thing to do is generate them all with %u.

9 years agoRemoved the offending debug message from code [Tue, 9 Aug 2011 16:22:59 +0000]
Removed the offending debug message from code

9 years agoRemoved the default small network (QEMU_NETWORK_SUPPORT) configuration option. [Tue, 9 Aug 2011 15:13:24 +0000]
Removed the default small network (QEMU_NETWORK_SUPPORT) configuration option.
Added net-test to x86_64 architecture

9 years agoApplying the diff with my tree as patch [Tue, 9 Aug 2011 15:09:35 +0000]
Applying the diff with my tree as patch

9 years agoFixed bug in the Glossary Haskell source (don't close stdout!).
Mothy [Mon, 8 Aug 2011 12:49:08 +0000]
Fixed bug in the Glossary Haskell source (don't close stdout!).
Added missing dependencies in Flounder interfaces in the distributed memory sever.
Fixed some make rehake bugs in

9 years agoKicking harness once more.
Simon Peter [Sat, 6 Aug 2011 07:15:49 +0000]
Kicking harness once more.

9 years agomackerel: fix minor bug that prevented changing the device prefix for the new shift...
Andrew Baumann [Fri, 5 Aug 2011 18:54:12 +0000]
mackerel: fix minor bug that prevented changing the device prefix for the new shift driver

9 years agoFixed spawn_program_on_all_cores() to check first whether a spawnd is
Simon Peter [Thu, 4 Aug 2011 14:32:00 +0000]
Fixed spawn_program_on_all_cores() to check first whether a spawnd is
running before trying to spawn a program on that core.

9 years agoDummy commit to kick harness.
Simon Peter [Wed, 3 Aug 2011 09:40:05 +0000]
Dummy commit to kick harness.

9 years agoFix tcpip.c.
Simon Peter [Wed, 3 Aug 2011 09:13:26 +0000]
Fix tcpip.c.

9 years agostack-rip send of intermon.bind_monitor_reply
Andrew Baumann [Tue, 2 Aug 2011 17:04:18 +0000]
stack-rip send of intermon.bind_monitor_reply

9 years agoRemoved Unicode characters from Mackerel can't handle them.
Simon Peter [Tue, 2 Aug 2011 15:51:24 +0000]
Removed Unicode characters from Mackerel can't handle them.

9 years agoGnaa! Shouldn't have used hg import to get the patch. Removing a lot of junk
Simon Peter [Tue, 2 Aug 2011 15:38:48 +0000]
Gnaa! Shouldn't have used hg import to get the patch. Removing a lot of junk
that shouldn't have gone in with the last commit.

9 years agoThese files are supposed to be deleted.
Simon Peter [Tue, 2 Aug 2011 15:23:11 +0000]
These files are supposed to be deleted.