November 2009 - Darwin-dev - lists.apple.com

Re: Alternative to bootstrap_register?
by Alastair Houghton 17 Nov '09

17 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha1; c=relaxed; d=alastairs-place.net; h= subject:mime-version:content-type:from:in-reply-to:date:cc: content-transfer-encoding:message-id:references:to; q=dns/txt; s= aug07; bh=HA14X7D5H0lQqwZHHP0hCKWA5vc=; b=NbPoZZQCE9w7VgoI0xdx8w ye5q1ZXWrSFBIHTgcbe0Jmzk45uy1BPDvk7r3p7x8akgWjZusK9YxyH4qrmWnpqV /scbaGJbaZ8IkLxIiz5vyja1F4y67+98MsN2zFBaFfPyvaVPawseekJWJGVHhpvK lXc34x0JnpXxIm3hjGEEY= On 17 Nov 2009, at 17:27, Uli Kusterer wrote: > In case of suggestion #3, the application that used to call bootstrap_register only has the pid of the target application to which to "send a Mach send-right directly". > > In order to send a Mach send right however one needs a send right itself to a port to which the target application has receive rights. In order to get that send right, the target application itself also needs to send this application a Mach send right, however in order send that send right it also needs a send right to send it and the loop never ends. > > Does anyone know what paragraph 3) means to achieve and how, possibly with a short example? :-) I think paragraph 3 is a bit bonkers, for the reason you give. You need a send right to a port in the target task in order to send a send right. Presumably the assumption is that someone may have *had* a send right already and decided to send a name via a separate route for some reason instead, though why you would do that is beyond me. If you only have a pid, you plainly can't send a send right. What you need to do is something like the example I posted on my blog the other day: <http://alastairs-place.net/2009/11/cross-process_s/> That is, use a server to register and name your send rights. As Tyler Sample pointed out to me today, you *could* use the bootstrap server for that purpose, but it's a little evil because you need to use bootstrap_register() (or something similar), and putting random ports in the bootstrap namespace seems a little evil to me. I prefer the separate server approach. Kind regards, Alastair. -- http://alastairs-place.net _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Alternative to bootstrap_register?
by Jean-Daniel Dupas 17 Nov '09

17 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Le 17 nov. 2009 à 18:27, Uli Kusterer a écrit : > In <servers/bootstrap.h> there is the following note about the bootstrap_register() call: > > * This API is deprecated. Old scenarios and recommendations: > * > * 1) Code that used to call bootstrap_check_in() and then bootstrap_register() > * can now always call bootstrap_check_in(). > * > * 2) If the code was registering a well known name, please switch to launchd. > * > * 3) If the code was registering a dynamically generated string and passing > * the string to other applications, please rewrite the code to send a Mach > * send-right directly. > > In case of suggestion #3, the application that used to call bootstrap_register only has the pid of the target application to which to "send a Mach send-right directly". > > In order to send a Mach send right however one needs a send right itself to a port to which the target application has receive rights. In order to get that send right, the target application itself also needs to send this application a Mach send right, however in order send that send right it also needs a send right to send it and the loop never ends. That's why the bootstrap server was invented ;-) > > Does anyone know what paragraph 3) means to achieve and how, possibly with a short example? > > To be noted, the open source CFMessagePortCreateLocal() still uses bootstrap_register years later after the API has been marked as deprecated. > Note really. It uses bootstrap_register2 which is a SPI and not deprecated. That said, this question reminds me a question I posted here 10 month ago :-) http://lists.apple.com/archives/darwin-dev/2009/Jan/msg00072.html Now it depends what you want to do. If you plan to register and checkin the service in the same process, it can be done using the NSMachBootstrapServer API. Now if you want to register the service, lookup, and wait that an other process checked-in, it's a little more complex. An approach may be to use a temporary launchd service. The new ServiceManagement Framework provide a way to dynamically register a launchd service (SMJobSubmit). You can register your service with "LaunchOnlyOnce" flag set so it will be automatically unregistred just after launch. If you need 10.5 compatibility, you can also use the plain launchd API. A little less friendly, but it works too. -- Jean-Daniel _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Alternative to bootstrap_register?
by Uli Kusterer 17 Nov '09

17 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In <servers/bootstrap.h> there is the following note about the bootstrap_register() call: * This API is deprecated. Old scenarios and recommendations: * * 1) Code that used to call bootstrap_check_in() and then bootstrap_register() * can now always call bootstrap_check_in(). * * 2) If the code was registering a well known name, please switch to launchd. * * 3) If the code was registering a dynamically generated string and passing * the string to other applications, please rewrite the code to send a Mach * send-right directly. In case of suggestion #3, the application that used to call bootstrap_register only has the pid of the target application to which to "send a Mach send-right directly". In order to send a Mach send right however one needs a send right itself to a port to which the target application has receive rights. In order to get that send right, the target application itself also needs to send this application a Mach send right, however in order send that send right it also needs a send right to send it and the loop never ends. Does anyone know what paragraph 3) means to achieve and how, possibly with a short example? To be noted, the open source CFMessagePortCreateLocal() still uses bootstrap_register years later after the API has been marked as deprecated. Cheers, -- Uli Kusterer "The witnesses of TeachText are everywhere..." _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Re: gcc compiler error: "Ignoring attempt to re-define symbol"
by Shantonu Sen 17 Nov '09

17 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Use -save-temps and look at the line in the assembly that it is complaining about, then find the other label of the same name, and determine why the label is present twice. Inline levels and inline asm can both contribute to this type of issue. Shantonu Sen ssen(a)apple.com Sent from my Mac Pro On Nov 16, 2009, at 4:10 PM, Giorgio Mazzucchelli wrote: > What I'm trying to do is to compile an open-source project under Mac OS; > in particular, there is no need to run any installation scripts, > to compile, only the gcc C compiler is required. > > Although the project supports both Linux and FreeBSD systems, > however, on the latter it runs into this kind of error: > > -Tmp-//ccYQ9y4o.s:1073:Ignoring attempt to re-define symbol. > make[2]: *** [obj_native/uaodv.o] Error 1 > make[2]: Target `all' not remade because of errors. > > The error couldn't be on the file's source code, > because the very same runs whit no problems on Linux (gcc version 4.2.4). > > > Any hints would be well-appreciated. > > Giorgio > > > Mac OS 10.5.8 > gcc version 4.2.1 (Apple build 5577) > Using built-in specs. > Target: i686-apple-darwin9 > > _______________________________________________ > 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/ssen%40apple.com > > This email sent to ssen(a)apple.com _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

gcc compiler error: "Ignoring attempt to re-define symbol"
by Giorgio Mazzucchelli 17 Nov '09

17 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:message-id:from:to :content-type:content-transfer-encoding:mime-version:subject:date :x-mailer; bh=PbuOql3Iyw+964pEtzMgZyb/AKYqisA0kAT4mI9GkOQ=; b=bJEliJGEXEb6zkgv+9zXkCFAEestMXJFdaUTa61MAbM3xyncWkjyFiT2iq44Cl22I7 mOMYJKcS4tzaZULs3hQJNXsr4PRqJal87htUDnovx24jc43wgOqb7FIj0fua2wqjA+j4 NwOXKFEGLmw6ToSaYNeFng2Sn5ccz2ApzttQE= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=message-id:from:to:content-type:content-transfer-encoding :mime-version:subject:date:x-mailer; b=PJvCh8YiV3833pUe8tePhwtBfWAD+D1gYlIzEZE+PpnKS2NxdczZzcn/o5zvaQPoGJ 9d/T6OQfreA7TTzCuD1WylPr2PfNen2838dVHCjai7uP6dzD7GcnaANxEqg4lZi0knuW lDgyES1QSoyt2XXGGjuvX/IDvP/jhHr9A+KDs= What I'm trying to do is to compile an open-source project under Mac OS; in particular, there is no need to run any installation scripts, to compile, only the gcc C compiler is required. Although the project supports both Linux and FreeBSD systems, however, on the latter it runs into this kind of error: -Tmp-//ccYQ9y4o.s:1073:Ignoring attempt to re-define symbol. make[2]: *** [obj_native/uaodv.o] Error 1 make[2]: Target `all' not remade because of errors. Any hints would be well-appreciated. Giorgio Mac OS 10.5.8 gcc version 4.2.1 (Apple build 5577) Using built-in specs. Target: i686-apple-darwin9 _______________________________________________ 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… The error couldn't be on the file's source code, because the very same runs whit no problems on Linux (gcc version 4.2.4). This email sent to site_archiver(a)lists.apple.com

1 0

rpcgen
by Chris Inacio 16 Nov '09

16 Nov '09

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:date:message-id:subject :from:to:content-type; bh=dJ89nSeu6cvwkP/OwMZtT324sHQCHxOqy9Z14rHPUfg=; b=NrJ1Az5CrgKhMlNVYkUij3ShtqtKwzZxEr/3pMDlPtiZKsCw/OZALFWbIuVWfLqxRb lAdYjESd1z45ZSe/LHXhSD0SEK1DLEL1fifPUOydnkeSC4w5DIdph4+4EVe63FkqrIjC q/wheT79Ld+82i/VZMAd7hF9hZrkWtQER/mZc= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:date:message-id:subject:from:to:content-type; b=VgslF8EOxTX7oleZOfirNZ5nBNFXj67R/OWPT8YluWFmHYZwFSqekunY3R8MQGRaPq 5VhwpinLj1qvohid1R7vV7tYETJ11YK3PiP1kJOlNH/MrOhW1LPYqQspq724hoGG1lgT dVU7F8+7/3a+kQ9OHYp2qu7Bb0i2LKLwtG/UQ= _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Blocks
by Jean-Daniel Dupas 16 Nov '09

16 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Blocks
by Damien Sorresso 16 Nov '09

16 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Nov 16, 2009, at 7:21 AM, Clark Cox wrote: > On Mon, Nov 16, 2009 at 2:51 AM, Filip van der Meeren > <filip(a)code2develop.com> wrote: >> Hi, >> >> Sorry if I am posting this on the wrong forum. But I really do not know >> where to post this message. >> Is it possible to archive/store a block? > > I do not believe so. Think about what that would entail; it would have > to save all of its code, as well as all of its context; and would have > to do so in a way that would remain valid across process launches > (what happens if someone archives a block in 32-bit code, but attempts > to call it from 64-bit code, etc.) I imagine that this is where using llvm to generate intermediary byte-code would be helpful. But preserving context would be the tricky part. If a block captures a pointer, that address won't be valid when you unarchive the block. You'd have to store what that address points to, load it into memory at unarchiving time and then update the block's captured pointer accordingly. So it gets really unwieldy when you start introducing things like CoreFoundation objects or Objective-C objects that don't comply with NSCoder. And it gets even worse if you're calling any API from within your block and assuming a certain state in the framework backing that API. The block has no idea what the framework's backing state is (maybe the framework needs to initialize a global, or certain things your program has done beforehand have changed that state from its initial value), so it can't capture that context and reinitialize. -- Damien Sorresso BSD Engineering Apple Inc. _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Blocks
by Clark Cox 16 Nov '09

16 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:in-reply-to:references :date:message-id:subject:from:to:cc:content-type; bh=JYyXu13ZE0ADgXXIdqd/juqmumS1boO/zJIRIFwZ2KU=; b=q4wWJl+hy2Wnuvh/c6Ulqe9x/bC9Y7I2dv7jWtx4L2kUhKqNyAA7sCYhgqzpcKxQ0M QPaxWNOtQF2zy6RNFoIfSEnQnl7wRCCaRujf0UZRZwN39G329zvvL96cEcIHICTRm2q4 kt3iSsnUiOCQRaNIsSuCtrfxnVoWeu0CV4qjo= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; b=sUCRdegc2rpJmeqecbPEWZS47LdDIqM0QSLpCSvN+2K0IOB8HRPd0H0o03tf3+XI9l uD+ZCH/qgnznDHaws0lw+hGH32kzl1iwmvX6wVqCBRDtRmxzKzKQ9OJHbZFcY4x1p3eH XguZV3EvGBbPTnloHn52EnEsSkkg9aBRThGgE= On Mon, Nov 16, 2009 at 2:51 AM, Filip van der Meeren <filip(a)code2develop.com> wrote: > Hi, > > Sorry if I am posting this on the wrong forum. But I really do not know > where to post this message. > Is it possible to archive/store a block? I do not believe so. Think about what that would entail; it would have to save all of its code, as well as all of its context; and would have to do so in a way that would remain valid across process launches (what happens if someone archives a block in 32-bit code, but attempts to call it from 64-bit code, etc.) -- Clark S. Cox III clarkcox3(a)gmail.com _______________________________________________ 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… This email sent to site_archiver(a)lists.apple.com

1 0

Blocks
by Filip van der Meeren 16 Nov '09

16 Nov '09

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi, Thanks, Filip _______________________________________________ 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… Sorry if I am posting this on the wrong forum. But I really do not know where to post this message. Is it possible to archive/store a block? This email sent to site_archiver(a)lists.apple.com

1 0