May 2005 - Darwin-dev - lists.apple.com

calling open() on an older cd device returns EBUSY??
by David M. Cotter 05 May '05

05 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thanks again! _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Thank you Shantonu so much for your help, your suggestion did indeed work, prepending an "r" to the device name. Apparently I didn't need to append "s1" to indicate session 1, or, rather, it seems to work just fine without that. Oh, and, list people, apologies for the "got any clue" thing below, I can see how that could be interpreted as, like "you have no clue" and I *very* much didn't intend to imply that, rather, I was feeling utterly clueless myself and am utterly grateful for the help this list is to all those with questions! This email sent to site_archiver(a)lists.apple.com

1 0

Re: How to know when memory pages are swapped?
by Steve Checkoway 05 May '05

05 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com - Steve _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… - mlock/munlock is available for user space code The caveat is that they must also be executable, to be wired from user space. Where is this documented? The man pages for mlock/munlock do not mention that at all. This email sent to site_archiver(a)lists.apple.com

1 0

Re: writing a new GUI environment
by Michael Buice 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On May 4, 2005, at 4:45 PM, darwin-dev-request(a)lists.apple.com wrote: Does anyone know how Apple's X implementation is done? It's not a Carbon or Cocoa app AFAIK... Michael A. Buice http://soma.uchicago.edu/~mabuice/blog _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I assume you mean the X11.app which displays X11 windows on Mac OS X. They subclassed the Cocoa NSApplication class as X11Application and started the event loop with an X11ApplicationMain() function which is similar to a standard NSApplicationMain() function, so in that sense I suppose it is a "Cocoa" application, as all the high level stuff is handled by Cocoa. The actual XQuartz server is started as a thread within this. I don't think this helps you any. I played around with the code for awhile to discover all of this for a project I was working on but my coding skills are more in numeric computation. You'll undoubtedly have better luck deciphering stuff: http://www.opensource.apple.com/darwinsource/tarballs/other/ X11ForMacOSXSource-1.0.tar.gz This email sent to site_archiver(a)lists.apple.com

1 0

Re: writing a new GUI environment
by Markus Hitter 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Am 04.05.2005 um 22:59 schrieb Erich Ocean: Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I'd like to use CGL *without* Carbon or Cocoa; I'm perfectly happy doing my own window management, so the full-screen thing would work. (It would be *nice* to be a full citizen in Mac OS X, but since this is for educational purposes, I can live with that.) What I'm concerned about is receiving keyboard and mouse events without having Carbon or Cocoa linked in. I'm getting the impression that that isn't possible. I'd guess a deep look into how X11 is implemented on / ported to Mac OS X would help you a lot. This email sent to site_archiver(a)lists.apple.com

1 0

Re: while (1) { dashboard();}
by robert delius royar 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Organization: An Apple OS X end user Priority: normal Thu, 5 May 2005 (08:03 +1000 UTC) Ian Lister wrote: Ian _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/x11%40frinabulax.org -- Dr. Robert Delius Royar Associate Professor of English Morehead State University Morehead, Kentucky Making meaning one message at a time. Organize a strike in your school or workplace on the grounds that it does not satisfy your need for indolence & spiritual beauty. -Hakim Bey 18:19 up 1 day, 9:15, 1 user, load averages: 1.14 0.94 0.72 _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Wed, 4 May 2005, Stephane Sudre wrote: aren't kern.maxprocperuid and kern.maxproc (I think it's these sysctl values) too small now that dashboard is in town? They've always been too small. I find it really easy to hit the limit in "normal" desktop use - I tend to have a lot of terminals open, and find it difficult to find some processes that I'm happy to terminate when I do hit the limit. I've had it in Radar for a while (#3537659) but last I heard there hadn't been any effort to address it. Thankfully the Darwin source releases have been very punctual for some time now (thanks Kevin!) so for each OS update I build a custom kernel with CHILD_MAX (xnu/bsd/sys/syslimits.h) increased from 100 to 500 and hack in a "ulimit -u" invocation in /etc/rc or somewhere (there being no login.conf or the like to be able to set it for login sessions). This email sent to x11(a)frinabulax.org Since 10.2 I have included the following at the bottom of /etc/rc.common. It works, or it has for me. if [ "${TWEAKS:=-NO-}" = "-YES-" ]; then ConsoleMessage "Tweaking stack, data, and processes" sysctl -w kern.maxproc=1024 sysctl -w kern.maxprocperuid=256 ulimit -H -u512 ulimit -u256 -s8192 -d12288 fi I tried adding it to rc, but the change wouldn't hold. Some late process in the load sequence would reset it (at least in 10.2). This email sent to site_archiver(a)lists.apple.com

1 0

Re: How to know when memory pages are swapped?
by Shaun Wexler 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On May 4, 2005, at 2:43 PM, Stéphane LETZ wrote: On 4 mai 05, at 23:24, Shaun Wexler wrote: Thanks a lot. Just to check again : _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I wrote a paged-buffer class with selectable features such as wired, virtual-ring, shared, etc. During starting of an audio device, I first wire down and touch each page of the I/O ring buffers; when the device stops, they are unwired and/or freed. Any buffers that my audio/DSP and MIDI time-constrain threads utilize, I wire when in use. There are a few MacFOH users who only have 128 to 384 MB of RAM installed, and nobody has reported any thrashing issues. - mlock/munlock is available for user space code The caveat is that they must also be executable, to be wired from user space. -- Shaun Wexler MacFOH http://www.macfoh.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: while (1) { dashboard();}
by Ian Lister 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Ian _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Wed, 4 May 2005, Stephane Sudre wrote: aren't kern.maxprocperuid and kern.maxproc (I think it's these sysctl values) too small now that dashboard is in town? They've always been too small. I find it really easy to hit the limit in "normal" desktop use - I tend to have a lot of terminals open, and find it difficult to find some processes that I'm happy to terminate when I do hit the limit. I've had it in Radar for a while (#3537659) but last I heard there hadn't been any effort to address it. Thankfully the Darwin source releases have been very punctual for some time now (thanks Kevin!) so for each OS update I build a custom kernel with CHILD_MAX (xnu/bsd/sys/syslimits.h) increased from 100 to 500 and hack in a "ulimit -u" invocation in /etc/rc or somewhere (there being no login.conf or the like to be able to set it for login sessions). This email sent to site_archiver(a)lists.apple.com

1 0

Re: writing a new GUI environment
by Shawn Erickson 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On May 4, 2005, at 2:28 PM, Shawn Erickson wrote: ...and note the link in the upper left corner... -Shawn _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… What I'm concerned about is receiving keyboard and mouse events without having Carbon or Cocoa linked in. I'm getting the impression that that isn't possible. Getting events would be interesting but you can likely do it without using Carbon or Cocoa, etc. but it would take work. Not sure myself how you would do it but interfacing with HID down closer to IOKit may do it. Yeah found what I was looking for... (I have used HID for joysticks but never tried for mouse and keyboards, left the later to Carbon or Cocoa) <http://developer.apple.com/documentation/DeviceDrivers/Conceptual/ HID/index.html> <http://developer.apple.com/documentation/DeviceDrivers/Conceptual/ HID/HID_Examples.dmg> This email sent to site_archiver(a)lists.apple.com

1 0

Re: writing a new GUI environment
by Shawn Erickson 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On May 4, 2005, at 1:36 PM, Erich Ocean wrote: #include <OpenGL/OpenGL.h> int main (int argc, const char * argv[]) { long value; long renders; CGLRendererInfoObj renderInfo; CGLQueryRendererInfo(1, &renderInfo, &renders); if (renders != 0) { printf("Render Info:\n\tnumberOfRenders = %ld\n", renders); CGLDestroyRendererInfo(renderInfo); } return 0; } [Session started at 2005-05-04 14:23:35 -0700.] Render Info: numberOfRenders = 3 Render 0x00022404 [0] : kCGLRPAccelerated = 1 kCGLRPFullScreen = 1 kCGLRPWindow = 1 kCGLRPBufferModes = 0x0000000f kCGLRPColorModes = 0x00008400 kCGLRPDepthModes = 0x00000c01 kCGLRPAccumModes = 0x00808000 kCGLRPStencilModes = 0x00000080 kCGLRPMaxSampleBuffers = 1 kCGLRPMaxSamples = 4 kCGLRPVideoMemory = 268435456 [256.00 MB] kCGLRPTextureMemory = 242024448 [230.81 MB] Render 0x00020200 [1] : kCGLRPAccelerated = 0 kCGLRPFullScreen = 0 kCGLRPWindow = 1 kCGLRPBufferModes = 0x0000000d kCGLRPColorModes = 0x0000ca00 kCGLRPDepthModes = 0x00001401 kCGLRPAccumModes = 0x00c0c000 kCGLRPStencilModes = 0x00000081 kCGLRPMaxSampleBuffers = 0 kCGLRPMaxSamples = 0 kCGLRPVideoMemory = 0 [0.00 MB] kCGLRPTextureMemory = 0 [0.00 MB] Render 0x00020400 [2] : kCGLRPAccelerated = 0 kCGLRPFullScreen = 0 kCGLRPWindow = 1 kCGLRPBufferModes = 0x0000000d kCGLRPColorModes = 0x0c0fc000 kCGLRPDepthModes = 0x00001000 kCGLRPAccumModes = 0x08808000 kCGLRPStencilModes = 0x00000081 kCGLRPMaxSampleBuffers = 1 kCGLRPMaxSamples = 16 kCGLRPVideoMemory = 0 [0.00 MB] kCGLRPTextureMemory = 0 [0.00 MB] CGLTest has exited with status 0. -Shawn _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On May 4, 2005, at 1:59 PM, Erich Ocean wrote: On May 4, 2005, at 1:46 PM, Shawn Erickson wrote: Mac OS X supports three (or perhaps four or five) distinct GUI enviroments: Carbon, Cocoa, Java, and perhaps QuickTime and OpenGL. OpenGL => NSGL, AGL, CGL or GLUT (<http://developer.apple.com/qa/ qa2001/qa1269.html>) CGL looks like it *should* be low-level, but in fact, it's only accessible via Carbon or Cocoa: No you can use it without linking against Carbon or Cocoa, just link against the OpenGL.framework. int i; for (i = 0; i < renders; i++) { CGLDescribeRenderer(renderInfo, i, kCGLRPRendererID, &value); printf("\tRender 0x%08x [%d] :\n", value, i); CGLDescribeRenderer(renderInfo, i, kCGLRPAccelerated, &value); printf("\t\tkCGLRPAccelerated = %d\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPFullScreen, &value); printf("\t\tkCGLRPFullScreen = %d\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPWindow, &value); printf("\t\tkCGLRPWindow = %d\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPBufferModes, &value); printf("\t\tkCGLRPBufferModes = 0x%08x\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPColorModes, &value); printf("\t\tkCGLRPColorModes = 0x%08x\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPDepthModes, &value); printf("\t\tkCGLRPDepthModes = 0x%08x\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPAccumModes, &value); printf("\t\tkCGLRPAccumModes = 0x%08x\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPStencilModes, &value); printf("\t\tkCGLRPStencilModes = 0x%08x\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPMaxSampleBuffers, &value); printf("\t\tkCGLRPMaxSampleBuffers = %d\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPMaxSamples, &value); printf("\t\tkCGLRPMaxSamples = %d\n", value); CGLDescribeRenderer(renderInfo, i, kCGLRPVideoMemory, &value); printf("\t\tkCGLRPVideoMemory = %d [%.2f MB]\n", value, value / (1024.0 * 1024.0)); CGLDescribeRenderer(renderInfo, i, kCGLRPTextureMemory, &value); printf("\t\tkCGLRPTextureMemory = %d [%.2f MB]\n", value, value / (1024.0 * 1024.0)); } If you want to take over the whole the screen then you could just use OpenGL (CGL) and Quartz and do all your own drawing. I'd like to use CGL *without* Carbon or Cocoa; I'm perfectly happy doing my own window management, so the full-screen thing would work. (It would be *nice* to be a full citizen in Mac OS X, but since this is for educational purposes, I can live with that.) You would use Quartz Services to understand, managed, and capture displays. <http://developer.apple.com/documentation/GraphicsImaging/Reference/ Quartz_Services_Ref> What I'm concerned about is receiving keyboard and mouse events without having Carbon or Cocoa linked in. I'm getting the impression that that isn't possible. Getting events would be interesting but you can likely do it without using Carbon or Cocoa, etc. but it would take work. Not sure myself how you would do it but interfacing with HID down closer to IOKit may do it. This email sent to site_archiver(a)lists.apple.com

1 0

Re: How to know when memory pages are swapped?
by Shaun Wexler 04 May '05

04 May '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com On May 4, 2005, at 2:04 PM, Godfrey van der Linden wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Yup this is basically the solution I was going to suggest. Keep a small pool of 'wired' memory around on the end of your I/O train. I'm afraid I'm not quite sure what the correct userland APIs are but I believe you are on the right track. You probably don't want to mlockall() in any case, but I'll guess p1003 is probably Posix 1003 spec (just a guess). I know that paging can be an issue, especially with audio, but an audio stream should be so hot that paging doesn't really occur. Have you been able to work out what pages are missing and why they are so cold? Perhaps a more complete analysis of page utilisation will reward you better then mlocking a bunch of pages? I wrote a paged-buffer class with selectable features such as wired, virtual-ring, shared, etc. During starting of an audio device, I first wire down and touch each page of the I/O ring buffers; when the device stops, they are unwired and/or freed. Any buffers that my audio/DSP and MIDI time-constrain threads utilize, I wire when in use. There are a few MacFOH users who only have 128 to 384 MB of RAM installed, and nobody has reported any thrashing issues. -- Shaun Wexler MacFOH http://www.macfoh.com This email sent to site_archiver(a)lists.apple.com

1 0