Marko Zec [Mon, 6 Apr 2009 22:29:41 +0000 (22:29 +0000)]
First pass at separating per-vnet initializer functions
from existing functions for initializing global state.
At this stage, the new per-vnet initializer functions are
directly called from the existing global initialization code,
which should in most cases result in compiler inlining those
new functions, hence yielding a near-zero functional change.
Modify the existing initializer functions which are invoked via
protosw, like ip_init() et. al., to allow them to be invoked
multiple times, i.e. per each vnet. Global state, if any,
is initialized only if such functions are called within the
context of vnet0, which will be determined via the
IS_DEFAULT_VNET(curvnet) check (currently always true).
While here, V_irtualize a few remaining global UMA zones
used by net/netinet/netipsec networking code. While it is
not yet clear to me or anybody else whether this is the right
thing to do, at this stage this makes the code more readable,
and makes it easier to track uncollected UMA-zone-backed
objects on vnet removal. In the long run, it's quite possible
that some form of shared use of UMA zone pools among multiple
vnets should be considered.
Bump __FreeBSD_version due to changes in layout of structs
vnet_ipfw, vnet_inet and vnet_net.
Add support for PowerPC kernel core files. This commit only adds
support for virtual core files (aka minidumps). physical core
files are not supported.
The implementation is cross-tool ready and can be used in a non-
powerpc hosted debugger to analyze PowerPC core files. It also
accepts core files that still have the dump header, as can be
the case within Juniper where TFTP-based kernel core files are
supported and savecore is not used to "extract" the core file
from some dump device.
John Baldwin [Mon, 6 Apr 2009 21:11:08 +0000 (21:11 +0000)]
When a stale file handle is encountered, purge all cached information about
an NFS node including the access and attribute caches. Previously the NFS
client only purged any name cache entries associated with the file.
PR: kern/123755
Submitted by: Jaakko Heinonen jh of saunalahti fi
Reported by: Timo Sirainen tss of iki fi
Reviewed by: rwatson, rmacklem
MFC after: 1 month
John Baldwin [Mon, 6 Apr 2009 19:12:47 +0000 (19:12 +0000)]
Change the default timeout for caching attributes of a directory in the NFS
client from 30 seconds to 3 seconds. After the recent changes to add
caching of negative name cache lookups, a negative name cache hit will
persist until the client notices the parent directory has changed. The
higher the attribute cache timeout on directories, the longer that can take,
so lower the default timeout for directories to match that of regular files.
Randall Stewart [Mon, 6 Apr 2009 14:27:28 +0000 (14:27 +0000)]
Ok, looking at the solution a bit closer, the level
calculation was too agressive. Instead we should only
look at each nibble. This makes it so we make
10.2.0.0 become 10.2/16 NOT 10.2/17.
Need to explore the non-cidr address issue. The two
may not be seperable..
David Schultz [Mon, 6 Apr 2009 13:50:04 +0000 (13:50 +0000)]
Return -1 instead of 0 upon reaching EOF. This is somewhat ill-advised
because it means getdelim() returns -1 for both error and EOF, and
never returns 0. However, this is what the original GNU implementation
does, and POSIX inherited the bug.
Ed Schouten [Mon, 6 Apr 2009 13:09:02 +0000 (13:09 +0000)]
Reduce the dcons polling frequency to 25 Hz.
It makes little sense to use 100 Hz polling in dcons. We cannot live
without polling, because that's just how dcons works. It polls the
buffer filled by the firewire hardware. 25 Hz is probably enough for
most use cases.
Robert Watson [Mon, 6 Apr 2009 10:32:40 +0000 (10:32 +0000)]
Add SDT DTrace probes for namei():
vfs:namei:lookup:entry takes parent directory vnode pointer, path to
look up, and lookup flags.
vfs:namei:lookup:return takes an error value, and if successful, the
returned vnode pointer.
Randall Stewart [Mon, 6 Apr 2009 10:09:20 +0000 (10:09 +0000)]
Class based addressing went out in the early 90's. Basically
if a entry is not route add -net xxx/bits then we should use
the addr (xxx) to establish the number of bits by looking at
the first non-zero bit. So if we enter
route add -net 10.1.1.0 10.1.3.5
this is the same as doing
route add -net 10.1.1.0/24
Since the 8th bit (zero counting) is set to 1 we set bits
to 32-8.
Users can of course still use the /x to change this behavior
or in cases where the network is in the trailing part
of the address, a "netmask" argument can be supplied to
override what is established from the interpretation of the
address itself. e.g:
route add -net 10.1.1.8 -netmask 0xff00ffff
should overide and place the proper CIDR mask in place.
Only raise WARNS to 6 on i386 and amd64, strict alignment platforms still
barf at some of the gratuitous pointer gymnastics, and I do not see a
simple solution.
Fix logic in MOD_LOAD handler to call dcons_attach after all successful
dcons_drv_init invocations. Testing return value for 0 does not work for
cases where dcons_drv_init was called already as part of low level
console initialization.
Andrew Thompson [Mon, 6 Apr 2009 00:22:49 +0000 (00:22 +0000)]
Provide a better commit log for r190735, forced by making a whitespace change.
Refactor how we interface with the root HUB. This is achieved by making a
direct call from usb2_do_request to the host controller for root hub requests,
this call will perform the controller specific register read/writes and return
the error code.
This cuts out a lot of code in the host controller files and saves one thread
per USB bus.
Ed Schouten [Sun, 5 Apr 2009 22:08:18 +0000 (22:08 +0000)]
Remove if_ppp(4) and if_sl(4).
Not only did these two drivers depend on IFF_NEEDSGIANT, they were
broken 7 months ago during the MPSAFE TTY import. if_ppp(4) has been
replaced by ppp(8). There is no replacement for if_sl(4).
If we see regressions in for example the ports tree, we should just use
__FreeBSD_version 800045 to check whether if_ppp(4) and if_sl(4) are
present. Version 800045 is used to denote the import of MPSAFE TTY.
Discussed with: rwatson, but also rwatson's IFF_NEEDSGIANT emails on the
lists.
Fix the build when KDB is disabled. The second instance of rfi in
trap_subr.S that is patched at runtime to rfid on 64-bit systems
is inside KDB-specific code, so don't patch it without KDB.
Add an Open Firmware access module for real-mode OF accesses to the PowerPC
build. This is required for the IBM Mambo simulator, as well as a variety
of non-Apple PowerPC hardware.
Andrew Thompson [Sun, 5 Apr 2009 18:21:51 +0000 (18:21 +0000)]
MFp4 //depot/projects/usb@159995
- add support for more complicated HID descriptors which can have multiple
definitions of the same field.
- remove old modulo patch in ums, which I think is due to bad HID parsing,
which should be fixed now.
Reported by: netchild
Submitted by: Hans Petter Selasky
Andrew Thompson [Sun, 5 Apr 2009 18:21:40 +0000 (18:21 +0000)]
MFp4 //depot/projects/usb@159992
at91_udp.c does not exist anymore, it is now replaced by at91dci in
src/sys/dev/usb/controller. Also remove the ohci_atmelarm.c because it is also
included in src/sys/conf/files
Andrew Thompson [Sun, 5 Apr 2009 18:21:21 +0000 (18:21 +0000)]
MFp4 //depot/projects/usb@159946
Some cancelable flags are always true. Substitute these away. These cancelable
flags were mostly useful with the root HUB which is now handled differently.
Andrew Thompson [Sun, 5 Apr 2009 18:20:38 +0000 (18:20 +0000)]
MFp4 //depot/projects/usb@159909
- make usb2_power_mask_t 16-bit
- remove "usb2_config_sub" structure from "usb2_config". To compensate for this
"usb2_config" has a new field called "usb_mode" which select for which mode
the current xfer entry is active. Options are: a) Device mode only b) Host
mode only (default-by-zero) c) Both modes. This change was scripted using
the following sed script: "s/\.mh\././g".
- the standard packet size table in "usb_transfer.c" is now a function, hence
the code for the function uses less memory than the table itself.
Andrew Thompson [Sun, 5 Apr 2009 18:20:03 +0000 (18:20 +0000)]
MFp4 //depot/projects/usb@159871
- bugfixes after the memory usage reduction patch
- Use "udev->pipes_max" instead of USB_EP_MAX
- Use correct "bmRequestType" for getting the config descriptor.
Fix KBI breakage by r190520 which affects older linux.ko binaries:
1) Move the new field (brand_note) to the end of the Brandinfo structure.
2) Add a new flag BI_BRAND_NOTE that indicates that the brand_note pointer
is valid.
3) Use the brand_note field if the flag BI_BRAND_NOTE is set and as old
modules won't have the flag set, so the new field brand_note would be
ignored.
Alan Cox [Sat, 4 Apr 2009 23:12:14 +0000 (23:12 +0000)]
Retire VM_PROT_READ_IS_EXEC. It was intended to be a micro-optimization,
but I see no benefit from it today.
VM_PROT_READ_IS_EXEC was only intended for use on processors that do not
distinguish between read and execute permission. On an mmap(2) or
mprotect(2), it automatically added execute permission if the caller
specified permissions included read permission. The hope was that this
would reduce the number of vm map entries needed to implement an address
space because there would be fewer neighboring vm map entries that differed
only in the presence or absence of VM_PROT_EXECUTE. (See vm/vm_mmap.c
revision 1.56.)
Today, I don't see any real applications that benefit from
VM_PROT_READ_IS_EXEC. In any case, vm map entries are now organized
as a self-adjusting binary search tree instead of an ordered list. So,
the need for coalescing vm map entries is not as great as it once was.
Perform a dummy stwcx. when we switch contexts. The context
being switched out may hold a reservation. The stwcx. will
clear the reservation. This is architecturally recommended.
The scenario this addresses is as follows:
1. Thread 1 performs a lwarx and as such holds a reservation.
2. Thread 1 gets switched out (before doing the matching
stwcx.) and thread 2 is switched in.
3. Thread 2 performs a stwcx. to the same reservation granule.
This will succeed because the processor has the reservation
even though thread 2 didn't do the lwarx.
Note that on some processors the address given the stwcx. is
not checked. On these processors the mere condition of having
a reservation would cause the stwcx. to succeed, irrespective
of whether the addresses are the same. The dummy stwcx. is
especially important for those processors.
Implement kernel core dump support for Book-E processors.
Both raw physical memory dumps and virtual minidumps are
supported. The default being minidumps.
Don't let trailing empty lines overwrite the result of the last command with 0.
This affects the built-ins eval, fc, and trap and also the string passed to sh
with the -c option.
Bruce M Simpson [Sat, 4 Apr 2009 15:32:23 +0000 (15:32 +0000)]
Fix mbuf chain layout pessimization:
in the case where a single mbuf is allocated due to
m_getcl() returning NULL, we already call MH_ALIGN,
so do not increment m->m_data in this case.
When removing or renaming snaphost, do not delve into request_cleanup().
The later may need blocks from the underlying device that belongs
to normal files, that should not be locked while snap lock is held.
Randall Stewart [Sat, 4 Apr 2009 11:43:32 +0000 (11:43 +0000)]
Many bug fixes (from the IETF hack-fest):
- PR-SCTP had major issues when skipping through a multi-part message.
o Did not look at socket buffer.
o Did not properly handle the reassmebly queue.
o The MARKED segments could interfere and un-skip a chunk causing
a problem with the proper FWD-TSN.
o No FR of FWD-TSN's was being done.
- NR-Sack code was basically disabled. It needed fixes that
never got into the real code.
- CMT code had issues when the two paths were NOT the same b/w. We
found a few small bugs, but also the critcal one here was not
dividing the rwnd amongst the paths.
Obtained from: Michael Tuexen and myself at the IETF hack-fest ;-)
PowerPC, meet kernel core dumps. The support is based
on a generic dumper that creates an ELF core file and
uses PMAP functions to scan and iterate over memory
chunks, as well as handle memory mappings used during
dumping.
the PMAP layer can choose to return physical memory
chunks or virtual memory chunks. For minidumps, the
chunks should be virtual.
The default MMU I/F implementation for the scan_md()
method returns NULL. Thus, when a PMAP implementation
does not implement the required methods, an empty
core file is created. Here, empty means having an ELF
header only.
The Serverworks SATA chipsets used in Apple G5 systems require requiring
the ATA status register with a 4-byte read request. This updates it, and
subsequent 1-byte reads will return the correct result.
This commit adds a hack to do this, which is currently ifdef'd powerpc,
although Linux and Darwin do this unconditionally on all platforms.
Add support for 64-bit PowerPC CPUs operating in the 64-bit bridge mode
provided, for example, on the PowerPC 970 (G5), as well as on related CPUs
like the POWER3 and POWER4.
This also adds support for various built-in hardware found on Apple G5
hardware (e.g. the IBM CPC925 northbridge).
Colin Percival [Fri, 3 Apr 2009 21:13:18 +0000 (21:13 +0000)]
Set SO_NOSIGPIPE on sockets used by phttpget. Without this, if
(1) phttpget is attempting to download enough files that it can't send
all the requests at once, and
(2) the remote server forcibly closes the connection, resulting in RST
packets being sent,
phttpget will receive a SIGPIPE and terminate without downloading all
of the files.
This is probably responsible for a number of hard-to-reproduce errors
with portsnap and freebsd-update.
Andrew Thompson [Fri, 3 Apr 2009 19:49:33 +0000 (19:49 +0000)]
Add interleaving root hold tokens from the CAM probe to disk_create and geom
provider tasting. This is needed for disk attachments that happen after threads
are running in the boot process.
Allow the NULL, RTLD_SELF and RTLD_NEXT handles to work with dlfunc(3).
dlfunc() called dlsym() to do the work, and dlsym() determines the dso
that originating the call by the return address. Due to this, dlfunc()
operated as if the caller is always the libc.
To fix this, move the dlfunc() to rtld, where it can call the internal
implementation of dlsym, and still correctly fetch return address.
Provide usual weak stub for the symbol from libc for static binaries.
dlfunc is put to FBSD_1.0 symver namespace in the ld.so export to
override dlfunc@FBSD_1.0 weak symbol, exported by libc.
Reported, analyzed and tested by: Tijl Coosemans <tijl ulyssis org>
PR: standards/133339
Reviewed by: kan
Sam Leffler [Fri, 3 Apr 2009 18:00:19 +0000 (18:00 +0000)]
o update dwds mcast handling after hoisting ieee80211_encap: frames need
to be encapsulated before dispatching to the driver
o eliminate M_WDS now that we call ieee80211_encap directly and can supply
the wds vap to indicate a 4-address frame should be created