gmane.linux.distributions.emc.devel

(no subject)

Chris Morley — 2008-11-21T23:18:46

------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/_______________________________________________ Emc-developers mailing list Emc-developers-5NWGOfrQmneRv+LV9MX5uipxlwaOVQ5f< at >public.gmane.org https://lists.sourceforge.net/lists/listinfo/emc-developers

Controlling EMC from a non-GPL application

Leslie Newell — 2008-11-19T18:42:47

I would like to write a plugin for SheetCam that provides a simple front end to EMC. This would integrate motion ontrol and CAM in one package. The problem is that SheetCam is closed source as I need to make a living out of it. Can this be done without violating the EMC license? If it can, is it possible to have two GUIs controlling EMC at the same time? For instance SheetCam could provide a basic position display and run/pause/stop controls. For anything more sophisticated, the user could use Axis or any other GUI. Les ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

coverity

Sebastian Kuzminsky — 2008-11-19T04:15:52

Some people have asked to see the emc2 Coverity results. Erinn at Coverity has generously offered to add more user accounts. If you email me your real name, what you want your Coverity login to be, and your email address, I'll forward it to them and they'll set it up.

hal_vti strangeness

Sebastian Kuzminsky — 2008-11-14T21:45:31

I'm going through the issues Coverity has found in our TRUNK, and while fixing a couple of bugs in hal_vti i noticed something strange. hal_vti.h defines MAX_CHANS to be 4, and that's used a lot in hal_vti.c. But hal_vti.c defines MAX_CHAN (without the S) to be 8, and uses that in one place in rtapi_app_main(). Looks like a bug to me, but maybe Alex Joni or someone who knows the code can check it out and verify.

the buildbot machine is ready

Sebastian Kuzminsky — 2008-11-14T18:17:16

John Kasunich and I talked about playing some more with Buildbot as a way to move the emc2 compile farm forward. We're not going to turn the existing compile farm off (at least not yet); we're going to stand up a parallel system and see if it's good enough to switch over to. I have set up a machine at to that end. It's running Ubuntu Server 8.04.1. The CU firewall exceptions just went through, so both its webserver and the buildmaster are accessible directly over the net. The old nappy URL is no longer needed. I'm running two buildslaves as KVM VMs on my laptop whenever i'm in network range, but it would sure help if other folks ran some slaves too. Buildslaves are easy and fun! All you need is a computer that can run python, check out and compile our tree, and make outgoing TCP connections. Slaves can be behind a NAT firewall. Setting up buildslaves is easy. You're welcome to have copies of my VMs, which will run anywhere with only minor tweaks. Or you can install your own by using something like this untested procedure (sorry about the linebreaks): # disable the screensaver or set it to only blank the screen (not soak up CPU cycles with fancy eye candy) # make a dedicated user (by convention i've been calling the user "farmer", and that's what this document calls it, but you can use any name you want) # do a cvs checkout as farmer (in a throw-away directory out of the way) to make sure sure you can, and to make sure you have the ssh hostkey of the CVS machine cvs -d :ext:anon-IvVUwROCvVnimY89xWQX+A< at >public.gmane.org:/cvs -z3 checkout -d /tmp/remove-this-dir emc2 # install emc2 build deps: libpth-dev, tetex-extra cd debian; ./configure sim dpkg-checkbuilddeps apt-get install build-essential $WHATEVER_ELSE_IS_NEEDED # let the farm user run "sudo make setuid" without a password by adding this line to /etc/sudoers: farmer ALL = ALL, NOPASSWD: /usr/bin/make setuid # install buildbot.deb, on dapper you need to uncomment universe in sources.list apt-get install buildbot # you get these three variables from the buildmaster admin # on dapper and probably other older versions of buildbot, the command is just "slave" instead of "create-slave" # a reasonable standard for build-slave names might be: "distro(-rtai)?-arch", for example "Ubuntu6.06-rtai-x86" or "Ubuntu8.04-x86-64" buildbot create-slave ~/BuildBot/slave/$SLAVE_NAME $MASTER $SLAVE_NAME $SLAVE_PASSWORD # edit buildbot.tac to say "usepty = 0" vi ~/BuildBot/slave/$SLAVE_NAME/buildbot.tac # do this whenever you want your build-slave to be running, or (on Ubuntu Hardy at least) add it to /etc/default/buildbot buildbot start ~/BuildBot/slave/$SLAVE_NAME Let's see how this goes!

switching spindle to rotary and viceversa

Arnoldo Riter CNC — 2008-11-13T23:53:16

Hello Everyone, Please le me introduce myself. I'm working with cnc machines from 2 years now, I been modifying machinery into cnc using steeper motor+PC via parallel port Since I found EMC I am using it for my new machines and I want to use the mesa 5i20+7i33+7i37 and Delta servodrives. I think that moving from steeper to servo won't be so easy but I'd like enjoy with it I'd like to switch the control of my spindle into a rotary axes, I'd use a one servo Delta 1000wts for the Spindle-Axis controlled by torque 0 to 10v I wouldn't like to use any extra hardware but only changing on the fly the way in which EMC “speaks” to the drivers How can I achieve that? It is posible to use and M code to change configuration of the Spindle to Axe and viceversa? Any information or clue will be welcome I wouldn't like to spam the forum, if this place is not for this kind of question please let me know that Thanks in advance Arnold ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

the threads.0 test

Sebastian Kuzminsky — 2008-11-13T20:40:26

The threads.0 test runs two threads, one named "fast" with a 100 us period and one named "slow" with a 1 ms period. The fast thread has a sampler component that samples threadtest.counter, which increases by one each run of the thread. The slow thread resets the counter so it goes back to 1 again. The current test verifies that the fast thread runs exactly ten times between each run of the slow one. This works well on actual hardware, but doesnt work on my KVM virtual machines. The scheduling is way off. But it's not a bug in EMC2 or RTAI, it's just how RTAI works on this laggy "hardware". Jeff said the same at 21:01:59 here: I propose we relax the test to verify only that both threads get run at all, by verifying that the sampler in the fast thread records a sequence of integers that starts at 1 and increases by one each time, or else gets reset back to 1. It's a simple thing to test and I have it working in my sandbox. Do you folks who actually know this stuff agree that this is the way to go?

Porting fix to 2.2 branch

Michael Buesch — 2008-11-12T23:12:15

Hi, Is there a chance of porting this trivial fix http://cvs.linuxcnc.org/cvs/emc2/src/hal/components/limit2.comp.diff?r1=1.5;r2=1.6 to v2_2_branch? I'm currently running 2.2-branch-cvs on a 2.6.26.8 kernel and it's a little bit inconvenient to keep this tiny compiletime fix around in the local checkout. Thanks :)

RELEASED: emc 2.2.7

Jeff Epler — 2008-11-12T01:47:16

I'm pleased to announce the next release of the emc 2.2 series, emc 2.2.7. This version contains many bugfixes, including many enhancements and bugfixes for the new "hostmot2" (hm2) family of drivers for the Mesa PCI and parport cards such as 5i20 and 7i33. For users of the hostmot2 family of drivers, some configuration changes have been unavoidable: * the hm2_5i20 driver has been renamed to hm2_pci to reflect the support for additional cards including 5i22 and 5i23. * stepgen timing parameters are "u32 ns" instead of "float seconds" * stepgens now have a ".enable" pin which must be connected to axis.#.amp-enable-out or other enabling signal The new packages are now available in the package repositories. If you have already installed an earlier 2.2.x version, this is available as an update in the Ubuntu update manager. If you are still using 2.1.7 but want to upgrade, instructions are here: http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?UpdatingTo2.2 For information on how to upgrade from 2.2.x to 2.2.7 without using the package manager, see http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl?InstallingUpdates CVS users can get this release by checking out the RELEASE_2_2_7 tag. Source tarballs are available from the package repository: http://linuxcnc.org/hardy/dists/hardy/emc2.2/source/emc2_2.2.7.tar.gz Many thanks to Seb Kuzminsky for his work on hostmot2 and particularly for meeting a deadline I set at the last minute. Thanks also to Alex Joni for building the hardy/i386 packages on short notice. Changes in this release: * pop up a message when the right kernel isn't running * halcmd will now show an error message instead of segfaulting when inadequate locked memory is available * allow emc2-sim to be built on 8.04 * stepconf: fix "negative SCALE" problem reported by Christopher Purcell * comp: fix bad C code generated from variable array declarations such as 'variable double output[8];' * documentation improvments * axis: "touch off" of rotary axis on metric machine * hostmot2: Support all the PCI, PC/104-Plus, and EPP "Anything I/O" boards: 5i20, 5i22, 5i23, 4i65, 4i68, and 7i43. PCI users should switch from hm2_5i20 to hm2_pci driver. * hostmot2 changes: Config changes required! pwmgen.scale now works as advertised; stepgen timing params are "u32 ns" instead of "float seconds"; and stepgens now have a .enable pin which needs setting. * hostmot2 feature additions: Added PDM mode for pwmgen; PWM and PDM frequency are now settable by the user; implemented stepgen maxaccel and maxvel; support open-drain GPIOs; all GPIOs now have some representation in HAL (though the ones used by module instances are restricted); encoder improvements. * Lots of hostmot2 bug fixes, mostly with pwmgen and stepgen. Thanks to all the brave testers! Jeff ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

[PATCH] Fix machine crash on kernel shutdown

Michael Buesch — 2008-11-07T10:20:46

This patch fixes a very hard to debug crash (took me about 5 hours) on an x86_64 kernel. It might not crash on an i386 kernel, because of the looser memory requirements. But in any case, it will access uninitialized memory. On x86_64 this will cause a machine crash on kernel shutdown. This is the oops that happens on an x86_64 kernel (without stacktrace, because it will otherwise scroll off-screen): http://bu3sch.de/misc/emc2-parport-crash.JPG The patch is a backport of code from cvs-trunk. So trunk is not affected by this issue. However, I suggest applying this to the next release (2.2.7?), because poking with uninitialized kernel memory is no good. This could even lead to crashes on i386, if somebody overwrites the memory (the struct device_driver) before shutdown. Index: emc2-2.2.6/src/hal/drivers/probe_parport.c =================================================================== --- emc2-2.2.6.orig/src/hal/drivers/probe_parport.c2008-11-07 10:58:39.000000000 +0100 +++ emc2-2.2.6/src/hal/drivers/probe_parport.c2008-11-07 11:02:32.000000000 +0100 < at >< at > -136,5 +136,7 < at >< at > } void rtapi_app_exit(void) { + if (pnp_registered_parport) + pnp_unregister_driver (&parport_pc_pnp_driver); hal_exit(comp_id); }

[PATCH] Fix emc2 on recent kernels

Michael Buesch — 2008-11-04T11:01:23

The following error happens when compiling emc2 on a recent kernel: hal/components/limit2.comp:14: error: expected identifier or ‘(’ before ‘{’ token hal/components/limit2.comp: In function ‘_’: hal/components/limit2.comp:22: warning: comparison of distinct pointer types lacks a cast hal/components/limit2.comp:22: warning: comparison of distinct pointer types lacks a cast make[2]: *** [/home/mb/emc2-cvs-trunk/src/objects/hal/components/limit2.o] Error 1 make[1]: *** [_module_/home/mb/emc2-cvs-trunk/src] Error 2 make: *** [modules] Error 2 This patch fixes emc2 compilation (stable and cvs trunk) on a recent kernel (tested 2.6.26.7). The kernel already defines a clamp() macro in linux/kernel.h. In case the macro is defined, use it. It's a typesafe macro. Somebody with commit access, please commit this. Thanks. Index: src/hal/components/limit2.comp =================================================================== --- src.orig/hal/components/limit2.comp2008-11-04 11:49:54.000000000 +0100 +++ src/hal/components/limit2.comp2008-11-04 11:51:09.000000000 +0100 < at >< at > -11,11 +11,14 < at >< at > typedef struct { float old_out; } limit2_data; +#ifndef clamp static inline double clamp(double v, double sub, double sup) { if(v < sub) return sub; if(v > sup) return sup; return v; } +#endif + FUNCTION(_) { double tmp = in; double maxdelta = maxv * fperiod;

moved hostmot2 firmwares to their own debianpackages

Sebastian Kuzminsky — 2008-11-01T20:45:59

Jeff suggested on the Wiki that the HostMot2 firmware images be split out into their own .debs: I added three new .debs: emc2-firmware, emc2-firmware-mesa-5i20-hostmot2, and emc2-firmware-mesa-7i43-hostmot2. emc2.deb Recommends emc2-firmware. This means emc2-firmware gets installed by default if the user installs emc2.deb via apt, but the user can uninstall emc2-firmware without breaking emc2.deb emc2-firmware installs no files of its own, but Recommends the per-board firmware .debs. Each of the two per-board firmware .debs installs all the .BITs and .PINs for that board. The end result is that users get all firmware by default, but they can uninstall the firmwares for any boards they dont want. Seem reasonable? If there's buy-off on this scheme, i'll add .debs for the other 4 Mesa hostmot2 boards.

spindle brake

Chris Morley — 2008-10-29T22:09:23

if some one programs m3 then m4 does EMC put the brake on for a certain amount of time, error, or just change directions of the spindle. On a related question when using constant surface feed does Emc wait till the spindle is up to speed or just assume that the spindle is up to speed. A guy wants to but the brake on to stop the spindle when he switches from m3 to m4 (apparently he doesn't program a m0) I guess i'm asking how EMC decides when the spindle brake goes on. I would think in most cases a brake should not be put on until the spindle is already stopped. but I don't see any pins for that either. Maybe I missed them.. Chris Morley _________________________________________________________________ ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

Tcp modbus - help with connection

Chris Morley — 2008-10-29T19:12:34

Hi guys I'm hoping for help. I want to test Classicladder's modbus slave and master over Ethernet (TCP). But I really have not much of a clue how to connect it all. networking for me has been connect the cables and let Ubuntu figure it out :) When loading classicladder I can specify a port number - default is 9502. What is this port number for? on the master I set the slave number by the IP number of the slave and can optionally set a port number (I assume 9502 is default) I tried this with a cross over cable direct between two computers and network manager seemed not to know anything about each other. I also tried this with a router between, Then of course I get IP numbers through network manager. Modbus Initializes, binds and then listens on the slave. It never returns from listening. I really don't know how to use the network tools (or if they are any help) to debug this. Suggestions ? Chris Morley _________________________________________________________________ ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

RTAI for realtime

Chris Morley — 2008-10-26T22:40:58

I have been reading a little about RTAI and Xenomai. I wonder Why EMC choose RTAI, Other then at the time Xenomai may not have been available (around 2005?). It seems that Xen is developed more or maybe more openly - I can find more info on it by far. Comparisons are rare but it seems that RTAI has possibly 10% less latency and Xen is more flexible in regards to platforms,architecture and debugging capability. If EMC and or HAL is to be used with anything else besides X86 could the realtime patch change? Is there any real intention to be able to use EMC/HAL with anything besides X86? Is the reason for RTAI over any other - lowest latency number, Why change what works, No interest, or a technical reason? I take it that EMC does not actively use RT-linux anymore because RTAI was better? Just curious really Chris Morley _________________________________________________________________ ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

realtime driver question

Chris Morley — 2008-10-23T20:38:24

Just wondering: When I made the driver for the Opto ac 5, I used Linux PCI driver calls writel() and readl () to write and read the ports. Was this bad (blocking) or slow for the realtime thread? I think there is a more direct way (using a pointer?) Can anyone comment? Thanks Chris Morley _________________________________________________________________ ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

[ emc-Bugs-2187024 ] Sherline support does not work

SourceForge.net — 2008-10-22T15:09:31

Bugs item #2187024, was opened at 2008-10-22 11:09 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=106744&aid=2187024&group_id=6744 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: Stepconf Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Hugo Boyer (hugomatic) Assigned to: Jeff Epler (jepler) Summary: Sherline support does not work Initial Comment: When using the "Sherline" option as the driver type, the resulting configuration does not work. Leaving the type to "Other" works. When setting the driver type to "Sherline", Stepconf should also set the parport pinout to "Sherline" and mark pins 1 and 10-17 as unused (and not supported by the Sherline driver), otherwise the generated configuration won't work. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=106744&aid=2187024&group_id=6744 ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

[ emc-Bugs-2184280 ] bad behavior after mdi duringmanual mode

SourceForge.net — 2008-10-21T12:08:25

Bugs item #2184280, was opened at 2008-10-21 07:08 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=106744&aid=2184280&group_id=6744 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: EMC2 Task Controller Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Jeff Epler (jepler) Assigned to: Nobody/Anonymous (nobody) Summary: bad behavior after mdi during manual mode Initial Comment: (emc2 TRUNK from around 10/21/08) It is possible to get task into a bad state by issuing an mdi while in manual mode How to reproduce: * Start configs/sim/axis.ini * F1 F2 Ctrl-home (machine on and homed) * Open a new terminal * change to configs/sim directory * run python Expected behavior: An operator message is displayed. subsequent jogs or mode switches work normally. Actual behavior: An operator message is displayed. "ERROR: need to be enabled, in coord mode for linear move". However, subsequent jogs do not work (it jogs only a very short distance) ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=106744&aid=2184280&group_id=6744 ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

pyVCP Error

John Thornton — 2008-10-21T12:01:29

When working through the manual in the pyVCP section and using halrun to test my panel I get an error any time the example has a in it. I have a file named tiny.xml with the following "This is a label:" ('Helvetica',20 When I run it with the line halrun loadusr pyvcp -c mypanel tiny.xml I get the following error Error evaluating xml file Widget label, Property font Syntax error: invalid syntax (, line 1) However, if I change the single quotes wrapping the font name to double quotes it all works. So I'm guessing a typo in the manual? John ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

[ emc-Feature Requests-2166870 ] Deep Cycle rigidpeck tapping.

SourceForge.net — 2008-10-14T21:53:04

Feature Requests item #2166870, was opened at 2008-10-14 22:53 Message generated for change (Tracker Item Submitted) made by Item Submitter You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=356744&aid=2166870&group_id=6744 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: None Group: None Status: Open Resolution: None Priority: 5 Private: No Submitted By: Robert (robert_h272) Assigned to: Nobody/Anonymous (nobody) Summary: Deep Cycle rigid peck tapping. Initial Comment: would be very nice to see a peck tapping cycle added into EMC for deep holes or that hard to tap metal. just adding a Q into the G33 cycle. so command could look like G33 Z-5.0 Q1.0 Kxx.xx (tap 5mm deep, with 1mm peck depth at XX feed, repeat untill we reach -5mm) where Q would denote the peck depth. work little like the G73 drilling where we just back off a little then continue on our way. then there should maybe be a 2nd cycle one to do a full retract back to the start level of the tapping cycle, maybe inabled with a R word? R0 or something G33 Z-5.0 Q1.0 R0 Kxx.xx (tap 5mm deep, 1mm deep peck/feed in retract back to clearance at XX feed, feed back into the hole. repeat until we get to -5mm) the last method i feel would not be used too often until pecking method with no full retract, above would add a huge advantage to EMCs rigid tapping. this saves a lot of tapping calls over and over to do a deep hole tapping routine. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=356744&aid=2166870&group_id=6744 ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/

User app not responding to sigint

Eric H. Johnson — 2008-10-10T12:56:21

Hi all, I think I must be missing something simple, but I don't see it. My LCD interface program emclcd (http://cvs.linuxcnc.org/cvs/emc2/src/emc/usr_intf/emclcd.cc?rev=1.5;content -type=text%2Fplain) isn't responding to the terminal interrupt when EMC shuts down. This part was copied right out of emcsh. emclcd is loaded with: loadusr emclcd In the source, the next to last line of main sets the signal and function to call with: signal(SIGINT, sigQuit); sigQuit is then defined as: static void sigQuit(int sig) { sockClose(sockfd); thisQuit(); } The socket to lcdproc is closed and thisQuit is called. thisQuit just does a bunch of clean up, however sigQuit is never getting called, thus emclcd continues to run after EMC has shut down. Does anything obvious stick out to anyone as to why this is never getting called? Thanks, Eric ------------------------------------------------------------------------- This SF.Net email is sponsored by the Moblin Your Move Developer's challenge Build the coolest Linux based applications with Moblin SDK & win great prizes Grand prize is a trip for two to an Open Source event anywhere in the world http://moblin-contest.org/redirect.php?banner_id=100&url=/