des [Tue, 27 Sep 2011 18:42:09 +0000 (18:42 +0000)]
Think first, commit second.
1. Allow the caller to select active mode.
2. Fix the envar logic so it *always* overrides the caller's flags.
3. Document the change from active to passive.
When script(1) reads EOF from input it starts spinning on zero-byte
reads eating 100% CPU. Fix this by skipping select on STDIN after
reading EOF -- permanently if STDIN is not terminal and for one second
if it is.
Also after reading EOF from STDIN we have to pass it to the program
being scripted. The previous approach was to write zero bytes into the
pseudo-terminal. This does not work because zero-byte write does not
have any effect on read. Fix this by sending VEOF instead.
Submitted by: Ronald Klop <ronald-freebsd8@klop.yi.org>
Discussed with: kib, Chris Torek <chris.torek@gmail.com>
Approved by: kib
MFC after: 1 week
This update eliminates a lock-order reversal warning discovered
whle tracking down the system hang reported in kern/160662 and
corrected in revision 225806. The LOR is not the cause of the system
hang and indeed cannot cause an actual deadlock. However, it can
be easily eliminated by defering the acquisition of a buflock until
after all the vnode locks have been acquired.
des [Tue, 27 Sep 2011 17:11:31 +0000 (17:11 +0000)]
Use fseeko() instead of fseek(). The rest of the code is off_t-aware,
but the use of fseek() means fetch(1) can't correctly resume a transfer
that was interrupted past the 2 GB mark.
des [Tue, 27 Sep 2011 15:57:13 +0000 (15:57 +0000)]
Followup to r225599: the fseek() was a no-op since the file was opened
in append mode. Open it in read-write mode instead. Also move the
fseek up one level to cover the (unlikely but not impossible) case where
the server accepts ranges but does not send a Content-Size header.
When a status pointer is passed in mfi_dcmd_command does not return an
errno (if the ioctl is successful), so move the test for NO_HW_PRESENT
outside of the error case. This should fix incorrect reporting for
systems with a dead or no battery.
Additionally, handle error codes other than NO_HW_PRESENT by omitting
the battery capacity display. LSI's supercap-based parts (CV series)
report their data using the same interface as battery-based parts,
except that they do not include the capacity stats (state of charge,
cumulative charge cycles, etc.)
Revert r225372:
wdog_kern_pat() acquires eventhandler mutex, thus it cannot work in
kernel context (from where kdb_trap() runs).
The right way to fix this is both offering the
cpu-stop-on-panic-and-skip-locking logic and also a context for KDB
to officially run. We can re-enable this (or a similar) improvement
when these 2 patches hit the tree.
Unbreak no-ip and no-inet6 module builds with ipfw. For now continue to
build the ip_fw_pfil.c hooks and ipfw even in case of no-ip under the
assumption that the private L2 hook (which hopefully eventually will be a
pfil hook as well) can still be useful.
Do not deliver SIGTRAP on exec as the normal signal, use ptracestop() on
syscall exit path. Otherwise, if SIGTRAP is ignored, that tdsendsignal()
do not want to deliver the signal, and debugger never get a notification
of exec.
Found and tested by: Anton Yuzhaninov <citrin citrin ru>
Discussed with: jhb
MFC after: 2 weeks
Prefer PJDLOG_ASSERT() and PJDLOG_ABORT() over assert() and abort().
pjdlog versions will log problem to syslog when application is running in
background.
- Add missing interdependencies to kerberos libraries. Some of the
kerberos libraries were not linked properly (missing dependencies),
which causes 3rd party applications linking to fail when --as-needed
ld flag is used. I also added the --no-undefined ld(1) flag to make
sure that there're no missing dependencies.
Update netfront so that it queries and honors published
back-end features.
sys/dev/xen/netfront/netfront.c:
o Add xn_query_features() which reads the XenStore and
records the TSO, LRO, and chained ring-request support
of the backend.
o Rename xn_configure_lro() to xn_configure_features() and
use this routine to manage the setup of TSO, LRO, and
checksum offload.
o In create_netdev(), initialize if_capabilities and
if_hwassist to the capabilities found on all backends.
Delegate configuration of if_capenable and the TSO flag
if if_hwassist to xn_configure_features().
Reported by: Hugo Silva (fix inspired by patch provided)
Approved by: re
MFC after: 1 week
Modify the netfront driver so it can successfully attach to
PV devices with the ioemu attribute set.
sys/dev/xen/netfront/netfront.c:
o If a mac address for the interface cannot be found
in the front-side XenStore tree, look for an entry
in the back-side tree. With ioemu devices, the
emulator does not populate the front side tree and
neither does Xend.
o Return an error rather than panic when an attach
attempt fails.
Reported by: Janne Snabb (fix inspired by patch provided)
PR: kern/154302
Approved by: re
Correct suspend/resume support in the Netfront driver.
Sponsored by: BQ Internet
sys/dev/xen/netfront/netfront.c:
o Implement netfront_suspend(), a specialized suspend
handler for the netfront driver. This routine simply
disables the carrier so the driver is idle during
system suspend processing.
o Fix a leak when re-initializing LRO during a link reset.
o In netif_release_tx_bufs(), when cleaning up the grant
references for our TX ring, use gnttab_end_foreign_access_ref
instead of attempting to grant the page again.
o In netif_release_tx_bufs(), we do not track mbufs associated
with mbuf chains, but instead just free each mbuf directly.
Use m_free(), not m_freem(), to avoid double frees of mbufs.
o Refactor some code to enhance clarity.
[ Forced commit. Actual changes accidentally included in r225704 ]
sys/dev/xen/control/control.c:
Fix locking violations in Xen HVM suspend processing
and have it perform similar actions to those performed
during an ACPI triggered suspend.
Sponsored by: BQ Internet
Approved by: re
MFC after: 1 week
Add suspend/resume support to the Xen blkfront driver.
Sponsored by: BQ Internet
sys/dev/xen/blkfront/block.h:
sys/dev/xen/blkfront/blkfront.c:
Remove now unused blkif_vdev_t from the blkfront soft.
sys/dev/xen/blkfront/blkfront.c:
o In blkfront_suspend(), indicate the desire to suspend
by changing the softc connected state to SUSPENDED, and
then wait for any I/O pending on the remote peer to
drain. Cancel suspend processing if I/O does not
drain within 30 seconds.
o Enable and update blkfront_resume(). Since I/O is
drained prior to the suspension of the VM, the complicated
recovery process performed by other Xen blkfront
implementations is avoided. We simply tear down the
connection to our old peer, and then re-connect.
o In blkif_initialize(), fix a resource leak and botched
return if we cannot allocate shadow memory for our
requests.
o In blkfront_backend_changed(), correct our response to
the XenbusStateInitialised state. This state indicates
that our backend peer has published sufficient data for
blkfront to publish ring information and other XenStore
data, not that a connection can occur. Blkfront now
will only perform connection processing in response to
the XenbusStateConnected state. This corrects an issue
where blkfront connected before the backend was ready
during resume processing.
Properly handle suspend/resume events in the Xen device
framework.
Sponsored by: BQ Internet
sys/xen/xenbus/xenbusb.c:
o In xenbusb_resume(), publish the state transition of the
resuming device into XenbusStateIntiailising so that the
remote peer can see it. Recording the state locally is
not sufficient to trigger a re-connect sequence.
o In xenbusb_resume(), defer new-bus resume processing until
after the remote peer's XenStore address has been updated.
The drivers may need to refer to this information during
resume processing.
sys/xen/xenbus/xenbusb_back.c:
sys/xen/xenbus/xenbusb_front.c:
Register xenbusb_resume() rather than bus_generic_resume()
as the handler for device_resume events.
sys/xen/xenstore/xenstore.c:
o Fix grammer in a comment.
o In xs_suspend(), pass suspend events on to the child
devices (e.g. xenbusb_front/back, that are attached
to the XenStore.
Restore the writing of the .bss sections of the dsos (not the main
executable) after r190885. The whole region for the dso is mmaped with
MAP_NOCORE flag, doing only mprotect(2) over .bss prevented it from
writing .bss to core files.
Revert the optimization of using mprotect(2) to establish .bss, overlap
the section with mmap(2).
Avoid starting the USB transfer if an error is already pending.
This change fixes a race in device side mode during clear-stall from
host, which can cause data to be sent too early on the given
endpoint.
Remove RA timer on an interface with !IFF_UP actively after starting to send
clean-up RA messages for shutting down. The RA timers could prevent the rtadvd
daemon from shutting down because ra_output() just ignored !IFF_UP interfaces
and TRANSITIVE->UNCONFIGURED state transition never happened due to it.
Copy ip6po_minmtu and ip6po_prefer_tempaddr in ip6_copypktopts(). This fixes
inconsistency when options are specified by both setsockopt() and ancillary
data types.
It is safe to initialize locks even on early boot (and it is the same
thing all the other architectures already do) thus just initialize
kernel_pmap in pmap_bootstrap().
Reported by: alc
Reviewed by: alc, marius
Tested by: flo, marius
Approved by: re (kib)
MFC after: 1 week
- Document $ipv6_cpe_wanif.
- Emphasize $ipv6_enable and $ipv6_prefer are deprecated.
- Add more detail descriptions about $ipv6_activate_all_interfaces.
- Add some more examples of $ifconfig_IF_ipv6.
- rtsold(8) and rtadvd(8) can be used even when ipv6_gateway_enable=NO now.
Fix a logic bug in pc-sysinstall creating partitions.
Improve exit when an error occurs.
Fix parsing to grab values which contain extra '=' signs.
Fix a bug setting the timezone properly.
Fix a usage bug when setting up with gmirror.
Allow a uzip file from local media to be used.
Allow specifying flags for "newfs" when using UFS as the file system.
Run custom commands after doing final cleanup / fstab generation
and such. Also fix using relative path for config file.
Fix long-standing thinko regarding maxproc accounting. Basically,
we were accounting the newly created process to its parent instead
of the child itself. This caused problems later, when the child
changed its credentials - the per-uid, per-jail etc counters were
not properly updated, because the maxproc counter in the child
process was 0.
Clear transmit checksum offload context state upon lem(4) interface
initialization. Prior to this change packets may be transmitted with an
incorrect checksum.
Em(4) already has an equivalent change in r213234.
Inquire about terminal type if starting from a serial console and do
not configure the host's networking if netbooted [1]. Also fix FTP
installations behind some firewalls [2].
Fix the enabling/disabling of Heartbeats and path MTU
discovery when using the SCTP_PEER_ADDR_PARAMS socket option.
Approved by: re
MFC after: 1 month.
In order to maximize the re-usability of kernel code in user space this
patch modifies makesyscalls.sh to prefix all of the non-compatibility
calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel
entry points and all places in the code that use them. It also
fixes an additional name space collision between the kernel function
psignal and the libc function of the same name by renaming the kernel
psignal kern_psignal(). By introducing this change now we will ease future
MFCs that change syscalls.
des [Thu, 15 Sep 2011 22:50:31 +0000 (22:50 +0000)]
When resuming an HTTP download, we failed to verify that the range
returned by the server matched what we requested, and blindly appended
what we received to what we already had. This could go two ways: if the
delivered offset was higher than expected, the local file would contain
duplicate data, while if it was lower than expected, there would be data
missing from the middle of the file. Furthermore, if the transfer was
interrupted again, each subsequent attempt would compound the error.
Fix the first problem by restarting the transfer from scratch if there
is a gap, and the second by explicitly seeking to the correct location
in the local file so as to overwrite any duplicated data.
PR: bin/117277
Approved by: re (kib)
MFC after: 3 weeks
Add IPv6 support to the ng_ipfw(4) [1]. Also add ifdefs to be able
build it with and without INET/INET6 support.
Submitted by: Alexander V. Chernikov <melifaro at yandex-team.ru> [1]
Tested by: Alexander V. Chernikov <melifaro at yandex-team.ru> [1]
Approved by: re (bz)
MFC after: 2 weeks
Rename the cc.4 and cc.9 modular congestion control related man pages to
mod_cc.4 and mod_cc.9 respectively to avoid any possible confusion with the cc.1
gcc man page. Update references to these man pages where required.
Requested by: Grenville Armitage
Approved by: re (kib)
MFC after: 3 days
Microoptimize the return path for the fast syscalls on amd64. Arrange
the code to have the fall-through path to follow the likely target.
Do not use intermediate register to reload user %rsp.
Ensure that 1-to-1 style SCTP sockets can only be connected once.
Allow implicit setup also for 1-to-1 style sockets as described
in the latest version of the socket API ID.
We are allocating some kilobytes of extra memory during USB device enumeration.
This does not change alot under FreeBSD, but makes sense for various embedded
operating systems using the FreeBSD USB stack, which have less memory
resources available.
Allow WPA PSKs up to 63 characters, the max length according to
wpa_supplicant.conf(5). Up to now, the PSK was truncated
to 15 characters (the length of the input field).
Include descriptions for keymaps missing from INDEX.keymaps. This fixes
the five keymaps referred to only by filename on the keymap selection
screen of bsdinstall.
Modify vfs_register() to use a hash calculation
on vfc_name to set vfc_typenum, so that vfc_typenum doesn't
change when file systems are loaded in different orders. This
keeps NFS file handles from changing, for file systems that
use vfc_typenum in their fsid. This change is controlled via
a loader.conf variable called vfs.typenumhash, since vfc_typenum
will change once when this is enabled. It defaults to 1 for
9.0, but will default to 0 when MFC'd to stable/8.