| Age | Commit message (Collapse) | Author |
|---|
|
rt and such; tested and ok miod drahn
|
|
for a bare FAT12/16/32 filesystem. If one is found, spoof it as 'i'.
This enables iPod shuffles and other umass devices that are shipped
with a bare FAT32 filesystem to be used as install media. And
generally makes things easier for the user just trying to mount some
media, e.g. floppies.
Does not make it safe to write a disklabel or MBR on such a device!
That will still obliterate the FAT filesystem.
Help and suggestions from tom@ and deraadt@. Feedback and suggestions
for future enhancements from espie@ mickey@ and peter@.
ok tom@ deraadt@
|
|
|
|
go out the wrong endpoint. Some debug massage.
|
|
were not sending the last data correctly before.
|
|
not attach properly until a delay was added. Also some debug quieting.
|
|
Limitations include failing to recieve packets that are a multiple of 64 bytes
and sending packets longer than 128 bytes appears to be messed up.
Work in progress.
|
|
|
|
the direction bit correctly. More cleanup.
|
|
|
|
|
|
right now that are supposed to be atomic with respect to interrupts and
SMP: atomic_setbits_int and atomic_clearbits_int.
All architectures other than i386 and amd64 get dummy implementations
since at first we'll be replacing operations that are done with
"a |= bit" and "a &= ~bit" today. More proper implementations will follow
kettenis@, miod@ ok
|
|
of the disk. Fixes the creation of spurious 'i' partitions on new iPod
Shuffle. And other devices with no MBR, just a FAT32 filesystem.
Pointed out most recently by Stephan A. Rickauer for his sixth
generation, clip-on iPod shuffle.
Fix tested by Stephan.
ok pedro@ tom@ miod@ 'nice' deraadt@
|
|
- LK_EXCLUPGRADE is never used.
- LK_REENABLE is never used.
- LK_SETRECURSE is never used. Because of this, the lk_recurselevel
field is always zero, so it can be removed to.
- the spinlock version (and LK_SPIN) is never used, since it was decided
to use different locking structure for MP-safe protection.
Tested by many
|
|
real need to be verbose here; discussed with drahn@
ok kettenis@, miod@, deraadt@ and others
|
|
in the swi instruction. elminates an illegal userland reference from the kernel
also elminates a bunch of code. This is a flag day for binaries from
before the middle of November.
|
|
statclock(), do not bother doing this in userret() anymore. As a result,
userret() does not need its pc and ticks arguments, simplify.
|
|
|
|
domain number such that we can distinguish between busses with the same bus
number that are behind different host bridges. Domains can be accessed by
using different device nodes.
ok deraadt@
|
|
likely that interrupts are properly counted in kern.cp_time sysctl. so top(1)
et al will show a value for time spent in interrupts.
tested by many, ok drahn@
|
|
errors to userland: make all cpu_setperf functions return void.
Tested by many, ok gwk@
|
|
|
|
for cpu_swapin() on hppa* which is kept).
|
|
should never be referenced outside the context of the process to which
this stack belongs unless we do the PHOLD/PRELE dance. Loads of code
doesn't follow the rules here. Instead of trying to track down all
offenders and fix this hairy situation, it makes much more sense
to not swap kernel stacks.
From art@, tested by many some time ago.
|
|
previous behaviour of just disabling the USB device controller)
|
|
for PXA27x/Zaurus, not enabled yet; for dlg and dale :)
|
|
DEBUG - this never happens anyway.
|
|
iso_disklabelspoof() and udf_disklabelspoof(). Use (msg &&
iso_disklabelspoof()) instead. Makes more arch's textually consistant.
No functional change.
|
|
extended partition except on alpha, hppa, hppa64 and mips64. So don't
spoof extended partitions as 'a' and try to read a disklabel from
them. Make all archs consistant.
When we can boot from extended partitions we can change to reading and
writing disklabels from them.
This means the first OpenBSD ('A6') extended partition will now be
ignored during spoofing, just as subsequent ones have always been.
Feedback from tom@, weingart@ and deraadt@.
|
|
partitions with 0 in the length field we don't need to test for
and skip partitions that are all zeros. Eliminating that code
allows elimination of a local pointer.
"0 length partitions are definately dead" deraadt@
|
|
De-register. Move declaration to top of function for DOS MBR
processing. Indent nit in macppc.
No functional change.
|
|
than lp->d_partitions[8+n++]. All other cases already use the pointer.
No functional change.
|
|
"The dosparts member of cpu_disklabel was not used for any persistant
data. Eliminate it, and just use the data being read in while
processing MBR and EBR records.
Should be no functional change.
ok weingart@ deraadt@"
|
|
ok jsg@
|
|
some architectures (kaboom); that will be fixed in the next iteration
hopefully!
|
|
data. Eliminate it, and just use the data being read in while
processing MBR and EBR records.
Should be no functional change.
ok weingart@ deraadt@
|
|
disklabel, we shouldn't write the disklabel in such partitions either.
Brings amd64, arm, i386, landisk, macppc, mvmeppc (which roll their
own MBR search in writedisklabel()) into line with alpha, hppa,
hppa64, mips64 (which use readdoslabel() in both reading and writing
the disklabel). If all goes well, soon everyone will be using
readdoslabel().
|
|
disklabel. Fixes 'double spoofing' of these partitions as both 'a' and
something in the 'i' to 'p' range.
OpenBSD changed its MBR partition type to 'A6' eight years ago. As a
backward compatibility measure NetBSD and FreeBSD MBR partitions were
still spoofed as 'a' partitions when no OpenBSD partition could be
found. This bit of backward compatibility is no longer required.
'we can get rid of it' deraadt@
|
|
|
|
don't include dkbad.h when nothing in it is used. Missed
arm/include/disklabel.h first time around.
|
|
|
|
r1.54 of i386/disksubr.c). The check incorrectly handled extended
partitions. Possibly a symptom of a deeper problem with extended
partition offsets that will be addressed post-4.0.
Problem reported/fix tested by A. Velichinsky, cause diagnosed by
otto@.
ok pedro@ mickey@ miod@ deraadt@
|
|
|
|
block past d_secperunit.
Fixes devices where bogus partitions with a starting offset >
d_secperunit were created. e.g. floppies formatted by Caldera DR-DOS.
ok pedro@
|
|
them return 'invalid geometry' when d_secpercyl == 0. While there move
the check to a consistant location (after the check of d_secperunit)
and use a consistant idiom (i.e. some readdisklabel()'s have no 'done'
label).
prodded by thib@ after a bad macppc experience. ok deraadt@
|
|
RAW_PART's p_size. Since d_secperunit is initialized to 0x1fffffff if
not specified there should be no need to use the magic number again.
And if d_secperunit was specified then that value should be used
instead of the magic number.
This was already being done for hp300, luna88k, mac68k, mvme68k,
mvme88k.
Should be a no-op.
|
|
ok drahn@
|
|
are held off for too long, allow the system to continue running, even
if time is lost.
|
|
BUS_DMASYNC_POSTREAD implemented otherwise the old device state
may still be present in the cache. Allows em(4) to work _much_ better on
iodata.
|
|
|
