src-test2/sys/alpha, branch release/6.3.0_cvsFreeBSD source treehttps://cgit-dev.freebsd.org/src-test2/atom?h=release%2F6.3.0_cvs2008-01-15T15:46:22ZThis commit was manufactured by cvs2svn to create tag2008-01-15T15:46:22Zcvs2svncvs2svn@FreeBSD.org2008-01-15T15:46:22Zurn:sha1:5cd221028ba5e7ada35a0dd196fb642b44f197d1
'RELENG_6_3_0_RELEASE'.
This commit was manufactured to restore the state of the 6.3-RELEASE image.
MF6: Add the workaround in busdma for busted PYXIS chipsets along with2007-12-10T20:16:07ZJohn Baldwinjhb@FreeBSD.org2007-12-10T20:16:07Zurn:sha1:edc0b087c36fc1dd2ad46d4d913716d6e052a367
the 'machdep.busdma_pyxis_bug' loader tunable.
Approved by: re (kensmith, blanket on sys/alpha)
MFC: Expand critical section coverage for ithread scheduling.2007-11-27T17:46:18ZJohn Baldwinjhb@FreeBSD.org2007-11-27T17:46:18Zurn:sha1:9addff1cb53c6cedb521a95e67c0023c247309ff
Approved by: re (kensmith)
MF6: Sync with i386 and expand support for bounce pages, etc.2007-11-27T17:45:59ZJohn Baldwinjhb@FreeBSD.org2007-11-27T17:45:59Zurn:sha1:a6c0a4ae535e6bdce771cf23072eb402bdf71e2b
Approved by: re (kensmith)
MFC: Remove mpte optimization from pmap_enter_quick().2007-11-05T16:18:00ZStephan Uphoffups@FreeBSD.org2007-11-05T16:18:00Zurn:sha1:e014671c76c2504ddad2fb42b0dca737cf826932
There is a race with the current locking scheme and removing
it should have no measurable performance impact.
This fixes page faults leading to panics in pmap_enter_quick_locked()
on amd64/i386.
Requested by: alc@
Fix panic on kernel startup. This allowed me to successfully run a2007-11-02T21:40:04ZWilko Bultewilko@FreeBSD.org2007-11-02T21:40:04Zurn:sha1:0cb46da0d877864bdc6acf98a6e2e9ebf64e3f15
6.3-PRE kernel.
Patch kindly provided by: jhb
MFC 1.39: Grrr, a fix for the rman_set_rid() stuff I missed in my earlier2007-10-24T12:36:25ZJohn Baldwinjhb@FreeBSD.org2007-10-24T12:36:25Zurn:sha1:60c98c8a09b9431bc9e59db4013d0d0abbcf8eb6
MFC.
MFC: Use rman_set_rid() to save the rid in new resources when they are2007-10-03T20:41:35ZJohn Baldwinjhb@FreeBSD.org2007-10-03T20:41:35Zurn:sha1:26d6b5e019247a415f459057fa623815e3f21527
allocated. This makes 'bus_free_resource()' safe to use in 6.x and also
fixes an issue with the recent PCI MSI code MFCs that caused them to never
update the MSI config registers.
Reported by: gallatin (MSI breakage)
Disable the module building of the linux bits on alpha and remove them2007-09-20T06:33:26ZAlexander Leidingernetchild@FreeBSD.org2007-09-20T06:33:26Zurn:sha1:1607c18075fb2ae478699cccdb647d38cb266498
from the corresponding NOTES. [1]
There's no corresponding linux_base in the Ports Collection since years,
and no linux distribution which is up-to-date and suitable (security,
widespread, ...) for us. The only complain regarding linux & alpha
we got so far is that the linuxulator doesn't build anymore. This
complaint came from wilko -- the alhpa-part of re@ -- only. As the
alpha platform is dead (no support from the manufacturer / linux
distributions) and deprecated (removed in -current) it was decided to
disable the linux bits from the build instead of trying to fix the
problems on alpha.
The alpha parts in the source are not removed so far, in case there are
1-2 people which actually still make use of this and are interested to
use the linuxulator (and fix the problems). Additionally this prevents
regressions on i386/amd64 when removing the alpha bits in shared code
(the people which care about those code parts don't have enough time ATM
to do extensive testing). This decission may be revisited at a later
point in time.
Discussed with: wilko [1]
sys/lock.h must precede sys/sx.h these days or the kernel won't build.2007-09-15T09:40:38ZWilko Bultewilko@FreeBSD.org2007-09-15T09:40:38Zurn:sha1:e7a466fb093f2574423ce7ec875bb160e7299f41
Looks like no-one builds Alpha kernels anymore?