[Bug 378] faxgetty loop due to modem error
A minor issue in Class 1 HDLC frame transmission can cause faxgetty to wait
forever in cases where an unexpected response (not OK or CONNECT) is seen
after transmitting MCF. Lee's fine patch (with tweaking by Patrice) fixes
this issue. Thanks Lee!
Darren Nickerson [Wed, 26 Mar 2003 05:00:03 +0000 (05:00 +0000)]
[Bug 393] lucent config shouldn't default with Class1MsgRecvHackCmd
With a lot more Lucent experience under our belts, it's clear that this
hack is actually only useful in a small proportion of lucent-based
setups. Thanks Lee!
Darren Nickerson [Wed, 26 Mar 2003 04:41:55 +0000 (04:41 +0000)]
[Bug 391] RingExtended broke some distinctive ring configurations
Change default RingExtended to null to prevent smashing distinctive
ring setups (causing HylaFAX to answer ALL calls). Thanks to Trevis
Becker for reporting the problem, and Lee Howard for the fix!
Darren Nickerson [Tue, 25 Mar 2003 04:32:29 +0000 (04:32 +0000)]
[Bug 394] Infinite loop in hfaxd leads to client/server lockup & 99% CPU consumption
Thanks to Aidan VanDyk and iFax Solutions for a fix to this uncommon, but
very severe lockup in client/server communication.
Darren Nickerson [Mon, 10 Feb 2003 22:41:11 +0000 (22:41 +0000)]
Issue: faxq receives QUIT (Q over FIFO) but takes forever to obey
This issue, identified and fixed by iFax Solutions and Aidan Van Dyk,
has plagued faxq for quite some time. In brief, faxq should postpone exiting
until all active jobs (at the time of the Q message) have been reaped in
order to capture the exit status of any sub-processes (preparing, sending
etc.). Unfortunately, the logic was not preventing new jobs (sleeping
or suspended) from becoming active, which could mean an endless supply
of "active" jobs on a busy server.
This patch corrects that bug, and should finally help "faxquit" do the
right things in a timely manner.
[Bug 379] adds NOTIFY_FAXMASTER option for FaxDispatch
Allow flexible control of notification to FaxMaster on inbound faxes.
Previous default would see FaxMaster being mailbombed ;-). Thanks Lee!
[Bug 383] Assertion failed in FaxQueuer
Remove the job from the active before trying to insert into the sleeping queue.
Thanks to iFax Solutions and Aidan Van Dyk for discovering and correcting this
issue.
[Bug 223] RFE: 2-D MMR
This milestone patch is the result of a lot of hard work, and Lee Howard
deserves to be congratulated for his perseverance in getting this to
work. Please try to test this latest addition as comprehensively as
possible if you have hardware capable of doing 2-D MMR.
[Bug 374] ModemRingResponse, CIDNumberAnswerLength, CIDNameAnswerLength
Patch developed by iFax Solutions which allows DID-DTMF data to be
delivered and received properly. This allows HylaFAX to support MultiTech's
MT5634ZBA-DID (an analog DID modem).
[Bug 373] bad shell code in bin/archive
Add a missing then" in an if-then-fi clause in the default archive script...
Thanks to John Miller for pointing this out!
[Bug 372] textfmt produces bogus Postscript code for formfeed coincident
with page break and single character files
If a formfeed is coincident with a page break generated by textfmt, the
Postscript code is incorrect. This patch corrects two problems, one
originally reported by Vu Pham and a second one noticed by Glenn. Thanks
to both of you, and to Lee for the patch reformat! -d
[Bug 371] mimencode vs. uuencode
Make MIME encoding of attached fax configurable via FaxDispatch. Useful
if MIMENCODE utility, whatever is chosen, happens to not have the ability
to generate Base64-encoded text. Thanks Lee!
[Bug 366] tiff resolution units interpretation should default as inches
Default to RESUNIT_INCH instead of RESUNIT_NONE to bring us in line with
TIFF spec. Also treat RESUNIT_NONE as PostScript units, which follows
libtiff behaviour. This work contributed by Chris Bainbridge, and filed
here by Lee. Thanks! -d
[Bug 343] increase RING timeout to 6 seconds
Applied Lee's patch to make this timeout configurable. Even 6 seconds was
not long enough in some installs as it turns out, so the configurability
saves a hack/recompile cycle at least ;-)
[Bug 318] sendfax can't send SUB or PWD specification
adds subaddressing capability for faxes queued with sendfax
Excellent teamwork from Lee and Patrice. Thanks guyz. -d
[Bug 267] Fix gcc warnings in -Wall mode
Aidan's second kill, which adds all const/non-const char*/u_char* operators
This required backing out Patrice's first fix to StackBuffer.h (id=324)
[Bug 267] Fix gcc warnings in -Wall mode
Fixes warning about using char * instead of const char * with fxStackBuffer
This is the first of Patrice's kills on the warning front.
[Bug 227] faxstate bad option output error
Applied patch to clean up garbage preceding Usage: info when an invalid
option (or no option) was specified. Thanks Patrice!
Darren Nickerson [Tue, 22 Oct 2002 01:05:41 +0000 (01:05 +0000)]
A race condition in the lockfile polling could lead to faxgetty looping
and consuming max CPU. This addresses the problem, and promotes HylaFAX
to a 4.1.5 release.
Darren Nickerson [Fri, 18 Oct 2002 19:08:28 +0000 (19:08 +0000)]
[Bug 358] FaxGetty missing action on the device (no LOCKWAIT and neccessary reset)
Somehow this patch, meant to be in 4.1.4, was missed.
Adds polling for lockfiles to faxgetty when in RUNNING state. Thanks Aidan!
[Bug 361] certain docq files are not removed by faxqclean
Now they are, thanks to a nice catch by Charly Baker and careful testing
by Lee. Thanks fellas!
[hylafax-devel] [Bug 358] FaxGetty missing action on the device
Now faxq sends faxgetty to LOCKWAIT after locking device for outgoing faxes,
so that the modem is certain to be reset after transmission
[Bug 359] Faxgetty segmentation fault on solaris with SunSAI/P serial device
This patch removes CONFIG_SELECT, and makes Dispatcher do a simple not NULL
check it's pointers before following them.
This means that if we ever have bit's set in the select fd_set's that we didn't
request, for whatever reason, we won't follow NULL pointers (and SEGFAULT).
Thanks Aidan!
Darren Nickerson [Fri, 20 Sep 2002 03:00:16 +0000 (03:00 +0000)]
[Bug 357] adds probemodem -c and -f options
-f prevents probemodem from using fuser (enabling one to run probemodem while
running faxgetty)
-c AT_COMMAND allows the user to query a single AT command rather than the
entire set
Thanks Lee!
Darren Nickerson [Sat, 14 Sep 2002 02:57:14 +0000 (02:57 +0000)]
[Bug 291] Solaris DSO support
Removed runtime activation of DSO support for Solaris, and backed out the
previous fix which hardwired DSO=no for Sun's Forte compiler, since the default
DSO=auto is safe once again. -d
Darren Nickerson [Fri, 13 Sep 2002 16:55:15 +0000 (16:55 +0000)]
[Bug 355] New: Class1ResponseWaitCmd
Fixes "DIS/DTC received 3 times" errors for some USRs when faxing
to some HP fax machines, in particular OfficeJets.
The fix actually comes from the efax package by Ed Casas with whom Lee has
corresponded. Thanks Lee & Ed!
Darren Nickerson [Fri, 13 Sep 2002 16:48:02 +0000 (16:48 +0000)]
[Bug 326] HylaFAX gets its own T.35 NSF manufacturer code
This patch adds those Japanese Fax Manufacturers to the NSF list. Note that
the translation was not done and/or was rough in a few spots. Also note that
if you intend to read the previously posted "Fax Manufacturer" TIFF, that you
must read the binary numbers right-to-left and not left-to-right as
traditional.
Fine work Lee - thanks!
Darren Nickerson [Fri, 13 Sep 2002 16:25:31 +0000 (16:25 +0000)]
[Bug 350] New: +Xmonth time multiplier in At syntax and December
When using +Xmonth and getting to December or the next year, the code would add
another month (it used a year of 11 months instead of 12 months).
Now it doesn't. Thanks Patrice!
Darren Nickerson [Fri, 13 Sep 2002 16:20:34 +0000 (16:20 +0000)]
[Bug 349] killtime couldn't be >= 24 days after sendtime
Setting killtime (LASTTIME) to a time >= 24 days after sendtime would not work
as expected. The number of days was passed through a modulo 24 so having a
killtime of now+31days resulted in a killtime of now+7 days.
Thanks to Patrice for this fix.
Darren Nickerson [Thu, 29 Aug 2002 14:10:55 +0000 (14:10 +0000)]
[Bug 346] New: adds ExtendedRing config option
This patch, submitted by Sascha Herrmann <sascha.herrmann@gmx.de>, allows the
extended RING identifier to be configured. Apparently Zyxel sends extended
RING information in the format... "RING;1234567890;12345". The default remains
the same, so existing configurations should encounter no problems.
Darren Nickerson [Tue, 27 Aug 2002 23:19:21 +0000 (23:19 +0000)]
[Bug 343] increase RING timeout to 6 seconds
At least one telco, which shall remain nameless but will probably be out
of business soon anyway, spaces RINGs by 5.1 seconds. This addresses the
problem counting rings with such a telco. Thanks Lee!
Darren Nickerson [Tue, 27 Aug 2002 22:56:42 +0000 (22:56 +0000)]
[Bug 342] adds config/lucent-isa
Adds explicit support for ISA Lucent-based board which appear to behave
differently than their PCI counterparts. Thanks Lee!
Darren Nickerson [Wed, 21 Aug 2002 16:01:07 +0000 (16:01 +0000)]
[Bug 339] avoid +FCERROR before page image data
This patch implements a Class1MsgRecvHackCmd option which allows the user to
send a command (i.e. "AT+FRS=1" or "<delay:3>") which should assist in ensuring
that the remote V.21 carrier is off before starting up our high-speed carrier
in reception. Thanks Lee!
Darren Nickerson [Thu, 15 Aug 2002 03:22:14 +0000 (03:22 +0000)]
[Bug 326] HylaFAX gets its own T.35 NSF manufacturer code
Added Lee's patch which adds recognition for all US and UK NSF codes.
Fine work Lee, both on the gathering and the patching.
[Bug 333] V34 update broke xferfaxstats and/or the actual logging mechanism
Patrice has supplied a fix for this inadvertent side effect of changing the
encoded parameters of a fax
Darren Nickerson [Tue, 23 Jul 2002 18:20:22 +0000 (18:20 +0000)]
[Bug 313] Generalize cover page
The original faxcover.ps, with SGI logo, is now an example file, while the
default faxcover.ps is now 100% generic. I messed with the makefile on this
one and someone should check my work. Thanks Aidan, Matthew & Lee!
Darren Nickerson [Sat, 20 Jul 2002 05:24:17 +0000 (05:24 +0000)]
[Bug 320] New: notify.awk can't deal with more than one ":" per line
The value of jobtag: can now contain colons. We just cloned the approach
taken with the status: param. Thanks Patrice!
[Bug 311] patch allows parsing of extended RING data
Simple patch developed by iFax Solutions which allows extended RING data
(common with EICON and Elsa modems) to be detected as a ring for for
ModemConfig::parseCID() - the function which uses CIDNumber and CIDName
config definitions - to actually parse the extended RING data string.
[Bug 312] faxgetty segfaults when receiving extremely long scan lines
The Patrice & Lee show have managed to make this potential denial-of-service
vulnerability a non-issue. Great work gentlemen!