causes data corruption in combination with certain bridges.
Information about this problem was kindly provided by davidch. [1]
- As BGE_FLAG_PCIX is meant to indicate that the controller is in
PCI-X mode, revert to the pre __FreeBSD_version 602101 method of
reading the bus mode register rather than checking the mere
existence of a PCI-X capability, which is also there when the
NIC f.e. is put into a 32-bit slot causing it not to be in PCI-X
mode. Setting BGE_FLAG_PCIX inappropriately could cause the NIC
to be tuned incorrectly.
PR: 128833 [1]
Reviewed by: jhb
MFC after: 3 days
When VLAN tagged frame is received the hardware sets 'LONG' bit of
Rx status word. It is always set when the size of received frame
exceeded 1518 bytes, including CRC. This VLAN tagged frame clears
'OK' bit of Rx status word such that driver should not rely on 'OK'
bit of Rx status word to pass the VLAN tagged frame to upper stack.
To fix the bug, don't use SIS_CMDSTS_PKT_OK for Rx error check and
introduce SIS_RXSTAT_ERROR macro that checks Rx errors. If we are
configured to accept VLAN tagged frames and the received frame size
is less than or equal to maximum allowed length of VLAN tagged
frame, clear 'LONG' bit of Rx status word before checking Rx
errors.
Reported by: Vladimir Ermako < samflanker <> gmail DOT com >
Tested by: Vladimir Ermako < samflanker <> gmail DOT com >
This change was erronously ommitted from the r185690, and attempt
to simply add the prototype to string.h has revealed that several
contributed programs defined local prototypes for strndup(), controlled
by autoconfed config.h. So, manually change #undef HAVE_STRNDUP to
#define HAVE_STRNDUP 1. Next import of the corresponding program would
regenerate config.h, overriding the changes in this commit.
No objections from: kan
to read-locking in the TCP input path, allowing greater TCP
input parallelism where multiple ithreads or ithread and netisr
are able to run in parallel. Previously, most TCP input paths
held a write lock on the global tcbinfo lock, effectively
serializing TCP input.
Before looking up the connection, acquire a write lock if a
potentially state-changing flag is set on the TCP segment header
(FIN, RST, SYN), and otherwise a read lock. We may later have
to upgrade to a write lock in certain cases (ACKs received by the
syncache or during TIMEWAIT) in order to support global state
transitions, but this is never required for steady-state packets.
Upgrading from a write lock to a read lock must be done as a
trylock operation to avoid deadlocks, and actually violates the
lock order as the tcbinfo lock preceeds the inpcb lock held at
the time of upgrade. If the trylock fails, we bump the refcount
on the inpcb, drop both locks, and re-acquire in-order. If
another thread has freed the connection while the locks are
dropped, we free the inpcb and repeat the lookup (this should
hardly ever or never happen in practice).
For now, maintain a number of new counters measuring how many
times various cases execute, and in particular whether various
optimistic assumptions about when read locks can be used, whether
upgrades are done using the fast path, and whether connections
close in practice in the above-described race, actually occur.
MFC after: 6 weeks
Discussed with: kmacy
Reviewed by: bz, gnn, kmacy
Tested by: kmacy
incremented using in_pcbref(), and decremented using in_pcbfree()
or inpcbrele(). Protocols using only current in_pcballoc() and
in_pcbfree() calls will see the same semantics, but it is now
possible for TCP to call in_pcbref() and in_pcbrele() to prevent
an inpcb from being freed when both tcbinfo and per-inpcb locks
are released. This makes it possible to safely transition from
holding only the inpcb lock to both tcbinfo and inpcb lock
without re-looking up a connection in the input path, timer
path, etc.
Notice that in_pcbrele() does not unlock the connection after
decrementing the refcount, if the connection remains, so that
the caller can continue to use it; in_pcbrele() returns a flag
indicating whether or not the inpcb pointer is still valid, and
in_pcbfee() is now a simple wrapper around in_pcbrele().
MFC after: 1 month
Discussed with: bz, kmacy
Reviewed by: bz, gnn, kmacy
Tested by: kmacy
message for r185765.
Noted by: rdivacky
Requested by: des
Commit message for r185765 should be:
In procfs map handler, and in linprocfs maps handler, do not call
vn_fullpath() while having vm map locked. This is done in anticipation
of the vop_vptocnp commit, that would make vn_fullpath sometime
acquire vnode lock.
Also, in linprocfs, maps handler already acquires vnode lock.
No objections from: des
MFC after: 2 week
sbuf instead of doing uiomove. This allows for reads from non-zero
offsets to work.
Patch is forward-ported des@' one, and was adopted to current code
by dchagin@ and me.
Reviewed by: des (linprocfs part)
PR: kern/101453
MFC after: 1 week
does final drop of the the dq reference to put it onto the free list.
There is a possibility that the dq would be found by another thread
after sync and before the dqh lock is acquired. If that other thread
drops the dq before we have taken the dqh lock, the dirty dq is put on
the free list.
Recheck the DQ_MOD after the dqh lock is relocked. Repeat dqsync() if
the dq is dirty. This ensures that up to date dq is written in the quota
file and fixes assertion in dqget().
Reported and tested by: Frode Nordahl <frode nordahl net>
MFC after: 3 days
Waiting for 1ms for each GMII register access looks overkill and it
may also decrease overall performance of driver because re(4)
invokes mii_tick for every hz.
Tested by: rpaulo
laptops. This includes battery presence detection, charging status, current
and voltage readouts, and charge level indication. The sysctl interface
is somewhat ACPI-like.
established a valid link or not. In miibus_statchg handler add a
check for established link is valid one for the controller(e.g.
1000baseT is not a valid link for fastethernet controllers.)
o Added a flag RE_FLAG_FASTETHER to mark fastethernet controllers.
o Added additional check to know whether we've really encountered
watchdog timeouts or missed Tx completion interrupts. This change
may help to track down the cause of watchdog timeouts.
o In interrupt handler, removed a check for link state change
interrupt. Not all controllers have the bit and re(4) did not
rely on the event for a long time. In addition, re(4) didn't
request the interrupt in RL_IMR register.
Tested by: rpaulo
drivers, there should be a 1us delay after every write when
bit-banging the MII. Also insert barriers in order to ensure
the intended ordering. These changes hopefully will solve the
bus wedging occasionally experienced with DM9102A since r182461.
- Deobfuscate dc_mii_readreg() a bit.
loader_conf_files="foo bar baz"
should cause loading the files listed, and then resume with the
remaining config files (from previous values of the variable).
Unfortunately, sometimes the line was ignored -- actually even
modifying the line in /boot/default/loader.conf sometimes doesn't work.
ANALYSIS: After much investigation, turned out to be a bug in the logic.
The existing code detected a new assignment by looking at the address
of the the variable containing the string. This only worked by pure
chance, i.e. if the new string is longer than the previous value
then the memory allocator may return a different address
to store the string hence triggering the detection.
SOLUTION: This commit contains a minimal change to fix the problem,
without altering too much the existing structure of the code.
However, as a step towards improving the quality and reliability of
this code, I have introduced a handful of one-line functions
(strget, strset, strfree, string= ) that could be used in dozens
of places in the existing code.
HOWEVER:
There is a much bigger problem here. Even though I am no Forth
expert (as most fellow src committers) I can tell that much of the
forth code (in support.4th at least) is in severe need of a
review/refactoring:
+ pieces of code are replicated multiple times instead of writing
functions (see e.g. set_module_*);
+ a lot of stale code (e.g. "structure" definitions for
preloaded_files, kernel_module, pnp stuff) which is not used
or at least belongs elsewhere.
The code bload is extremely bad as the loader runs with very small
memory constraints, and we already hit the limit once (see
http://svn.freebsd.org/viewvc/base?view=revision&revision=185132
Reducing the footprint of the forth files is critical.
+ two different styles of coding, one using pure stack functions
(maybe beautiful but surely highly unreadable), one using
high level mechanisms to give names to arguments and local
variables (which leads to readable code).
Note that this code is used by default by all FreeBSD installations,
so the fragility and the code bloat are extremely damaging.
I will try to work fixing the three items above, but if others have
time, please have a look at these issues.
MFC after: 4 weeks
multiple algorithms and potentially collect multiple samples.
Instead of a single calibration interval we now have short and long
intervals; the long interval roughly corresponds to the previous
single interval. The short interval is used to speedup collection
of samples and happens much quicker. We make calls using the short
interval until we're told the calibration work is complete at which
point we fallback to the long interval. In addition there is a
much longer reset interval used to flush all calibration state and
cause everthing to start anew.
With these changes you can also disable calibration entirely by
setting the long interval to zero.
at. I don't think this will make a huge difference, but I have
received a report of a interrupt storm on one 16-bit card that this
might fix (chances are it won't, since I think that we may need to
check both the CBB registers for the 16-bit card as well as the PCIC
registers for power state change).
Submitted by: jhb@
the power interrupt and init code waiting for the interrupt are
running on different CPUs. I haven't seen this make any real
difference, but I've also had some reports of odd behavior I can't
otherwise explain. It is an infrequent operation, and certainly
wouldn't hurt.
