- LOCKPARENT is no longer relevant for lookup(), relookup() or VOP_LOOKUP().
these now always return the parent vnode locked. namei() works as before.
lookup() and various other paths no longer acquire vnode locks in the
wrong order via vrele(). fixes PR 32535.
as a nice side effect, path lookup is also up to 25% faster.
- the above allows us to get rid of PDIRUNLOCK.
- also get rid of WANTPARENT (just use LOCKPARENT and unlock it).
- remove an assumption in layer_node_find() that all file systems implement
a recursive VOP_LOCK() (unionfs doesn't).
- require that all file systems supply vfs_vptofh and vfs_fhtovp routines.
fill in eopnotsupp() for file systems that don't support being exported
and remove the checks for NULL. (layerfs calls these without checking.)
- in union_lookup1(), don't change refcounts in the ISDOTDOT case, just
adjust which vnode is locked. fixes PR 33374.
- apply fixes for ufs_rename() from ufs_vnops.c rev. 1.61 to ext2fs_rename().
routing caused by stale route caches (struct route). Route caches
are sprinkled throughout PCBs, the IP fast-forwarding table, and
IP tunnel interfaces (gre, gif, stf).
Stale IPv6 and ISO route caches will be treated by separate patches.
Thank you to Christoph Badura for suggesting the general approach
to invalidating route caches that I take here.
Here are the details:
Add hooks to struct domain for tracking and for invalidating each
domain's route caches: dom_rtcache, dom_rtflush, and dom_rtflushall.
Introduce helper subroutines, rtflush(ro) for invalidating a route
cache, rtflushall(family) for invalidating all route caches in a
routing domain, and rtcache(ro) for notifying the domain of a new
cached route.
Chain together all IPv4 route caches where ro_rt != NULL. Provide
in_rtcache() for adding a route to the chain. Provide in_rtflush()
and in_rtflushall() for invalidating IPv4 route caches. In
in_rtflush(), set ro_rt to NULL, and remove the route from the
chain. In in_rtflushall(), walk the chain and remove every route
cache.
In rtrequest1(), call rtflushall() to invalidate route caches when
a route is added.
In gif(4), discard the workaround for stale caches that involves
expiring them every so often.
Replace the pattern 'RTFREE(ro->ro_rt); ro->ro_rt = NULL;' with a
call to rtflush(ro).
Update ipflow_fastforward() and all other users of route caches so
that they expect a cached route, ro->ro_rt, to turn to NULL.
Take care when moving a 'struct route' to rtflush() the source and
to rtcache() the destination.
In domain initializers, use .dom_xxx tags.
KNF here and there.
* Add another model that needs to be added into the quirk table.
* Don't iterate over the pnow_cpu_quirk array if the two signatures
are identical.
Patch sent privately by Olaf 'Rhialto' Seibert, thanks.
Formac ProNitron 80.IVb (1024x768/8)
Relax 19" Model 200 (1024x768/8)
Apple Monochrome Video Card (640x480/1)
VillageTronic Mac Picasso 320 (up to 1920x1080/up to 24)
Nubus identifiers for Creative Solutions parallel & serial cards,
and a Becton Dickinson data acquisition card.
- add a ses(4) major number as the MI MAKEDEV tries to create the
nodes. hi elad.
- create audio devices.
- provide mount_kernfs and a dmesg command.
- start sysinst automagically and fix backspace key.
Mostly prompted by David H. Gutteridge on port-hp700.
processes through sysctl
Fix inverted logic in boolean assignment. This is why these tests should
not be done outside the secmodel code.
Thanks for the report.
let one create a tunnel with equal inner and outer destination IP
numbers. Update gre(4) documentation for this change.
Extract subroutine gre_update_route() from gre_compute_route(),
and always call it in gre_output() to freshen the route for
tunnel-encapsulated packets.
No functional change intended.
Add some new diagnostic code, bracketed by #ifdef RN_DEBUG, that
uses the two new subroutines to walk and print a tree.
XXX The format of the diagnostic print-outs needs improvement.
Probably it ought to come from some generic .h file, but I've no idea
which - nor if any existing define has the same effect.
In any case this saves oodles of bytes in bootxx_msdos.
> - Don't set MIIF_NOISOLATE so rgephy(4) can be used in configurations
> with multiple PHYs and un-comment case IFM_NONE in case MII_MEDIACHG
> rgephy_service(). There doesn't seem to be a problem with isolating
> RTL8169S and their internal PHY.
and rev 1.12:
> - Fix some spelling in comments.
> - Remove superfluous returns at the end of void functions.
> - Remove unused static global rgephy_mii_model.
- remove #ifdef'ed out code whcih has also been removed in FreeBSD
- some style cosmetics
in6_control() with splnet()/splx(). I was being a bit paranoid
here. Following a cursory analysis of the code, this still looked
necessary. We don't spend a lot of time in these calls, so it
should not be too harmful to suspend network interrupts.
In in6_unlink_ifa(), call in6_delmulti() just once on each multicast
address (in6_multi). Previously, in6_unlink_ifa() called in6_delmulti()
on each in6_multi until in6_delmulti() removed the in6_multi from
the list and freed its memory. That's not justified: the multicast
list holds *one* reference. All other references belong to other
entities. We must wait to free the memory until the other entities
release their references, to protect against dereferencing a freed
in6_multi.
XXX I need to revisit in6_delmulti(), in6_unlink_ifa(), and friends,
XXX to pry apart the conditions where an in6_multi is removed from
XXX its list and where it is freed. Following my change, above,
XXX we still risk dereferencing a freed in6_multi.
Prevent in6_update_ifa() and in6_addremloop() from creating dangling
pointers to interfaces in the routing table. Previously, my NetBSD
tunnel concentrator, which adds and deletes a lot of P2P interfaces
with the same local address, crashed in 8 hours or less when it
dereferenced a dangling pointer to a deleted ifnet. Now, its uptime
is greater than 3 days.
than allocating memory, and it does wrongly use the hub's capabilities
but not the actual setting
-switch a high-speed hub to "multiple TTs" but ignore errors; since
we don't care whether there is one or multiple this is a "best effort"
thing
corruption for incoming netiso packets with recent (at least NetBSD-3 and
later) compilers. This is done in a way that the copy is avoided totally.
Code path tested with tcp+udp/ipv4+ipv6, arp and ISO cltp/clnp.
Visually ok'd by Christos@.
nonsense quirk that switched operating mode on ICH7 and ICH8.
I removed the obvious candidates for ahcisata(4), and I'll have a closer
look later if there are others to be removed; ahcisata(4) will take over
handling the device anyway, but there is no reason to keep AHCI devices in
that list.
Along the way, remove the code that tries to put the chip in Enhanced mode,
it makes absolutely no sense to do that, and some BIOSes might not have
prepared the BARs for that, as proven by PR#34885. If people want to use
all IDE and SATA channels, they have to tell the BIOS.
handle a single entry and veriexec_table_delete() to handle an entire
table. veriexec_convert() now takes a struct vnode *, and made
veriexec_table_lookup() take struct mount * (that's entirely internal
now).
Tested on amd64, built successfully on amd64, i386, sparc, and sparc64.
poll interval: since this is a highspeed device, it should be a "12"
(just for sanity; this stuff is not used by the kernel, only userland
usb exploration code might see it)
From the PR:
Since struct scsipi_read_discinfo_data is an odd number of bytes long
(35 bytes), ti ends up being unaligned. wdc_datain_pio() uses
bus_space_read_multi_4() to read the response from the drive, and
since ti is unaligned, it causes an unaligned access fault.
Add __aligned__(4) to all the structures.
even if we've detected a 'root on raid' autoconfigure. This change is really
only cosmetic, since setroot() will still do the right thing and honor
the 'root on foo' setting.
unlocked instead of locked. chuq agrees
* use slock set to &uobj->vmobjlock also for the last simple lock
operation to be consistent with the rest of the function
RDSK block, check for a valid native NetBSD label. If found, use the
NetBSD label. When writedisklabel() gets called, if no RDB was found,
write out the native NetBSD label. The disk won't be usable by AmigaDOS,
but is usable as a 'normal' NetBSD drive, with partition changes saved in
the disk label.
for work on some future functionality.
- Veriexec data-structures are no longer exposed.
- Thanks to using proplib for data passing now, the interface
changes further to accomodate that.
Introduce four new functions. First, veriexec_file_add(), to add
a new file to be monitored by Veriexec, to replace both
veriexec_load() and veriexec_hashadd(). veriexec_table_add(), to
replace veriexec_newtable(), will be used to optimize hash table
size (during preload), and finally, veriexec_convert(), to convert
an internal entry to one userland can read.
- Introduce veriexec_unmountchk(), to enforce Veriexec unmount
policy. This cleans up a bit of code in kern/vfs_syscalls.c.
- Rename veriexec_tblfind() with veriexec_table_lookup(), and make
it static. More functions that became static: veriexec_fp_cmp(),
veriexec_fp_calc().
- veriexec_verify() no longer returns the entry as well, but just
sets a boolean indicating whether an entry was found or not.
- veriexec_purge() now takes a struct vnode *.
- veriexec_add_fp_name() was merged into veriexec_add_fp_ops(), that
changed its name to veriexec_fpops_add(). veriexec_find_ops() was
also renamed to veriexec_fpops_lookup().
Also on the fp-ops front, the three function types used to initialize,
update, and finalize a hash context were renamed to
veriexec_fpop_init_t, veriexec_fpop_update_t, and veriexec_fpop_final_t
respectively.
- Introduce a new malloc(9) type, M_VERIEXEC, and use it instead of
M_TEMP, so we can tell exactly how much memory is used by Veriexec.
- And, most importantly, whitespace and indentation nits.
Built successfuly for amd64, i386, sparc, and sparc64. Tested on amd64.
the d_typename field of the default disklabel is initialized with
text from the SCSI inquiry, so that one gets at least a vague idea which
disk he is dealing with
to ensure trap code will work (that is 0 upto ${endkernel}) and leave the
rest to pmap_tlbmiss(). Mapping whole physmem into the kernel wired way
too many TLB entries, see
http://mail-index.netbsd.org/port-powerpc/2006/10/27/0000.html
for performance analysis. While there, be a bit more descriptive in
pmap_tlbmiss() comment and use macro instead of numeric constant.
OK by Simon Burge
consinit() was moved before oea_init() in rev 1.136, but startsym and
endsym are not initialized at that point.
Now kernel symbols are usable (on ddb(4), ksyms(4) etc.) again on macppc.
It causes a few second TRX stalls on each ioctls
(on ifconfig(8) or tcpdump(8) etc.) and PHY seems
initialized properly without it at least on RTL8169S
with internal rgephy(4) PHY. Closes PR kern/34517.
XXX: maybe we should also test this on the original RTL8169
with external makphy(4).
- set RGEPHY_1000CTL_AHD in rge_reset()
- reset RGEPHY_MII_BMCR register in rge_reset() too
While here, use proper macro rather than magic numbers for
ANAR and BMCR registers there.
Annotate a memory leak.
When copying one multicast address list to another, IFAREF before IFAFREE
to protect against using an ifaddr after (accidentally) freeing it.
LIST_REMOVE() a multicast address from its old list before
LIST_INSERT_HEAD() on its new list.
Do not count on in6_delmulti() removing its multicast-record argument
from the multicast address list that the record belongs to, because
clearly that is not what it (always) does.
Obviously, this breaks the already unstable Veriexec ABI, but that's
it. Some cool additions are planned to be introduced, and this just
makes it so that NetBSD 4.0 users will be able to easily use them as
well.
This also removes the fingerprint type name limit, so relevant code
was adjusted.
Thoroughly tested (even uncovered a bug in proplib! thanks for fixing
that cube@!). Documentation updated.
transfer. It was wrong because it didn't consider block boundaries.
Instead, use the updated toggle from the transfer descriptor -- this
is supposedly cheaper than modulo arithmetics.
This fixes a problem similar to PR kern/34070 for me.
First, remove process_checkioperm() calls from MD code. Similar checks
using kauth(9) routines (on the process scope, using appropriate action)
are done in the callers.
Add secmodel back-end to handle each subsystem.
Revision 1.9:
> Make the rgephy driver advertise both 1000_FD and 1000_HD bits in
> autoneg mode. A couple of the devices don't autoneg correctly unless
> configured this way.
Revision 1.11:
> Due to the poor PHY documentation from RealTek I can't sure but I
> think the RealTek PHY needs driver to set RGEPHY_BMCR_AUTOEN bit of
> RGEPHY_MII_BMCR register and proper ANAR register setting for manual
> media type selection.
> This fixes long standing manual media type selection bug in rgephy(4).
using both the rawio_spec and passthru actions to detect raw disk
activity. Same for kernel memory policy.
Update documentation (no longer need to expose veriexec_rawchk()) and
remove all Veriexec-related bits from specfs.
* XXX: This is bogus. We should be failing the request
* XXX: not only if this specific slice is mounted, but
* XXX: if it's on a disk with any other mounted slice.
- checking if dmamap != NULL is not valid because dmamap is not cleared
in bus_dmamap_unload(9)
- no need check RX mbufs and call m_freem() and bus_dmamap_unload()
in vge_init() since it's done in prior vge_stop()
Now vge(4) works fine on Ultra5.
arithmetic on ILP32 (sizeof(bus_addr_t) == 4, exactly) hosts
- prepare and use VGE_PREV_TXDESC() macro
- use VGE_[TR]XDESCSYNC() calls more efficiently
- wrap a sanity check against VGE_RDSTS_OWN in vge_newbuf()
with #ifdef DIAGNOSTIC since it should not happen
- use sc->sc_tx_free (number of free TX descs) to check if TX packets are
queued or sent
- call vge_start() only if the interupt is actually handled by this driver
- some misc cosmetics
length and define VGE_RX_PAD (which is 4 bytes) for ETHER_ALIGN (2 bytes)
padding only in !__NO_STRICT_ALIGNMENT case to avoid confusion.
As per comments from Murata Shuuichirou in PR kern/31323.
Tested on i386 and macppc.
still enabled, as it could have been disabled through USERCONF.
Use it in amd64 and i386 mainbus code and skip all ACPI processing in case
it is disabled.
In OpenBSD's if_bgereg.h, CHIPID 0x4101000 is defined as BCM5750_B1
but our PR kern/31028 says it's BCM5751_A1 on BCM5751M on IBM T43p,
and the value seems reasonable.
Use Request Sense only if Read Capacity succeeded and did not return
valid block size.
Discussed on tech-kern.
Fix the easier part of NetBSD PR kern/26537.
(The harder part is the device hangs on large (>= 8KB) transfer.
Possibly umass BBB problem?)
Remove scsipi_size() and scsipi_validate_secsize() from scsipi_base.c
and add their functions to sd.c since they are used only by sd.c.
Use SCSI term `block' instead of `sector' where applicable.
macallan