January 2010 - Darwin-dev - lists.apple.com

Re: Kextload bus error on PowerPC
by Brian Bechtel 07 Jan '10

07 Jan '10

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:references:message-id:from:to :in-reply-to:content-type:content-transfer-encoding:x-mailer :mime-version:subject:date:cc; bh=l3Ybq20AxaQby4TrD0Mil7sal7W+8wW4KgLYhjkaTIo=; b=rs0Evl9zjos4MHC37iEpYWnfK/DYo4XhzcxRQBe6KOTxlb9K2ssCRF4rT3dN+rTPAy yu0wRSjodTZ01ZOsF/ufKA3U/TQ5OmUx/kR1Ve+gN3E+rHs50WGhWbWezENpvzTmFYmJ NVKZzLa8dW+PzMfGF1w+BbuJDI5VfjfuFIQd8= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=references:message-id:from:to:in-reply-to:content-type :content-transfer-encoding:x-mailer:mime-version:subject:date:cc; b=YYQCUHNf/zYmgh1NTS+V56wGkLllhPzvwSQVjJLyogm+tCZnztGb/0ZeZEqMJXPb9D BlE+2LOQtFvGyDDVGCY3o+lL2fqCeVNG9T4Tt7tyR/1L88XBAXibixO/d+oV6q6dyYA4 lBXZR66u7Mxq6X6aMyxuC8g8IAkc7nmZtCJQE= * this isn't magic. You have a bus error because of code in your kext Good luck. _______________________________________________ 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… You haven't given us specific info to diagnose this. Here are some points. * ppc machines completely ignore x86_64 architecture. The presence or absence of such code in your kext is not relevant. * you should use two machine debugging to diagnose this. You can use an intel machine to connect to your ppc panic; just specify the -arch keyword when invoking gdb * since you claim the i386 code works, I would suspect endian issues in your code. Look at Apple'sdiscussions in the universal programming guidelines for ideas This email sent to site_archiver(a)lists.apple.com

1 0

Kextload bus error on PowerPC
by Ryan McGann 07 Jan '10

07 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com I am working with an IOKit kernel extension (OSBundleRequired == "Root") that is built on Snow Leopard/Intel, but is targeting 10.4, 10.5 and 10.6 on PowerPC, i386 and x86_64. The i386 version of the kext loads and works fine, but I am having a problem with my PowerPC kernel extension (Tiger or Leopard). The machine panics pretty quickly after startup. When I boot in safe mode and load the kext manually, kextload complains "Executable does not contain kernel extension code". I figured out this was due to the presence of the x86_64 slice in the binary, so I tried building without that slice. kextload then says that the kernel extension can be loaded, but I just get a bus error. I tried all the usual suspects—turning off all stripping, removing the LC_UUID load command using strip, and more. No avail—the i386 version works fine on all platforms, so I can't figure out what the difference is between ppc and i386. Thanks, Ryan _______________________________________________ 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: appending to a memory-mapped file
by Jens Alfke 06 Jan '10

06 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jan 6, 2010, at 1:39 PM, Jean-Daniel Dupas wrote: > Just an idea, you can try to remap the same file with a larger size at the same address than the previous one using the MAP_FIXED flag. According to the man page, if you pass this flag, mmap try to "remove a mapping that already exists at that location". So it will automatically invalidate your previous mapping and create a new one with the new range (if there is enough room to do this of course, else it will failed). It doesn't fail if there's not enough room; it just wipes out any other mappings that overlapped the address range. When I tried doing this, it ended up overwriting a large malloc zone, which caused some exciting and hard-to-debug crashes. :-( —Jens _______________________________________________ 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: appending to a memory-mapped file
by Jean-Daniel Dupas 06 Jan '10

06 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Le 6 janv. 2010 à 17:36, Jens Alfke a écrit : > > On Jan 6, 2010, at 6:11 AM, Joel Reymont wrote: > >> Forgive me if I'm missing something obvious but how do you append to a memory-mapped file? >> The file will obviously grow when you append to it but what's the best way to 'extend' the memory representation? > > I've looked into this too. A memory-map applies not to the entire file, but to a range of bytes; and there's no way to modify one in place. You have to create a new mapping using mmap on the byte-range you added to the file. In general this means the maps will be discontiguous in address space. If that's not acceptable, you can try to reserve more address space than you need, use the beginning of it for the file map, and then position your extended maps after the first one (until you run out of space...) > > I'm not an expert, just quoting from memory. I haven't tried to do the contiguous-allocation stuff myself. Look in the list archives, either in this list or darwin-userlevel, circa Oct/Nov, for my questions and the answers I got. Just an idea, you can try to remap the same file with a larger size at the same address than the previous one using the MAP_FIXED flag. According to the man page, if you pass this flag, mmap try to "remove a mapping that already exists at that location". So it will automatically invalidate your previous mapping and create a new one with the new range (if there is enough room to do this of course, else it will failed). And it is the same file, the memory content should not change. I don't know if the OS try to optimize the operation by detecting that this is the same fd, and preserve the memory in this case. -- 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

Re: appending to a memory-mapped file
by Jens Alfke 06 Jan '10

06 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jan 6, 2010, at 6:11 AM, Joel Reymont wrote: > Forgive me if I'm missing something obvious but how do you append to a memory-mapped file? > The file will obviously grow when you append to it but what's the best way to 'extend' the memory representation? I've looked into this too. A memory-map applies not to the entire file, but to a range of bytes; and there's no way to modify one in place. You have to create a new mapping using mmap on the byte-range you added to the file. In general this means the maps will be discontiguous in address space. If that's not acceptable, you can try to reserve more address space than you need, use the beginning of it for the file map, and then position your extended maps after the first one (until you run out of space...) I'm not an expert, just quoting from memory. I haven't tried to do the contiguous-allocation stuff myself. Look in the list archives, either in this list or darwin-userlevel, circa Oct/Nov, for my questions and the answers I got. —Jens _______________________________________________ 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: appending to a memory-mapped file
by Jean-Daniel Dupas 06 Jan '10

06 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Le 6 janv. 2010 à 15:16, Jean-Daniel Dupas a écrit : > Le 6 janv. 2010 à 15:11, Joel Reymont a écrit : > >> Forgive me if I'm missing something obvious but how do you append to a memory-mapped file? >> >> The file will obviously grow when you append to it but what's the best way to 'extend' the memory representation? >> > > ftruncate passing the mapped file descriptor should work. Sorry, I answer too fast. I don't think it will increase mapped region, only the file size. -- 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

Re: appending to a memory-mapped file
by Jean-Daniel Dupas 06 Jan '10

06 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Le 6 janv. 2010 à 15:11, Joel Reymont a écrit : > Forgive me if I'm missing something obvious but how do you append to a memory-mapped file? > > The file will obviously grow when you append to it but what's the best way to 'extend' the memory representation? > ftruncate passing the mapped file descriptor should work. -- 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

appending to a memory-mapped file
by Joel Reymont 06 Jan '10

06 Jan '10

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:from:content-type :content-transfer-encoding:subject:date:message-id:to:mime-version :x-mailer; bh=3xErlQeKLl00aYjrGbDLD/7/zNAqFMcYbPWEXo/8oS0=; b=sNHCSI5pbXBlET7QCyLjudm3pPYbtLEXoNKZCs3nCLAgyVjQv426ZxSav0ypk2E58o EoDReRHcvNjmGCG/NAZrwGnE2SYbpF6+xkdwyN3W2wV297rnGgAlH7Ip9EsnG9dBmIgO pBnwI8379j1s6fdl2GP2SUm7oQrtFCwTGkReQ= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=from:content-type:content-transfer-encoding:subject:date:message-id :to:mime-version:x-mailer; b=oflwuHUtfcosC1KSe6aJAQTaB6QFPuYRDZhC+ifTDEnf5i0gqY/DI+fRt7n84auUwT vaMcGjkhsfLMO66MF60FkVaC+gEKrKfu0pPvG9ntsgN3cKngEgqPAVdaf+J6RgrvY9fv GhA/6de6WyfEFZs7cNTyPWDhZKMW0XOWw5iAo= Forgive me if I'm missing something obvious but how do you append to a memory-mapped file? The file will obviously grow when you append to it but what's the best way to 'extend' the memory representation? Thanks, Joel --- http://wagerlabs.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: IOLog printing to system message buffer instead of /var/log/system.log?
by Eric Gouriou 06 Jan '10

06 Jan '10

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha1; c=relaxed; d=pobox.com; h=subject :mime-version:content-type:from:in-reply-to:date:cc :content-transfer-encoding:message-id:references:to; s=sasl; bh= bTWTp2wURYpQROUJRPmk+Ka64Vw=; b=Nx9/wcGns2RxDcXjnWmG6PJrn74O/XkT lfBxcO8An9topLA37RBYwhzXsEScvVQTXFb7dU1FcVk4CgKRMgteNdrJLIYI9bOI aJh1KaRTR6fp35ThF7560/UAkPpvgXRs9cH9bnioHfetN9ZpuGWhsQu8p8TDgDm+ MweUlMat8zo= Domainkey-signature: a=rsa-sha1; c=nofws; d=pobox.com; h=subject :mime-version:content-type:from:in-reply-to:date:cc :content-transfer-encoding:message-id:references:to; q=dns; s= sasl; b=X0ALRd2/e74UxqYpPl3u0cO6U5IcEMSUVl10P1nVvd9VUDXLzVwbNcDe B/k4IVKoPH7ORTRJqrGg38Lte9nmxJJLW7r4OBvll+ZACnCASuxOMjhuky45g8qL cHv4u69qOQpVNKvz4Mx8WdkFOuw+DqudD9rVIy47UinSUb6UdOQ= On Jan 5, 2010, at 5:15 PM, Danny Mardini wrote: > Hi everyone, > > I'm sure there must be a simple answer to this question, I sincerely apologize if I shouldn't be asking it here. > > I'm creating a IOKit driver which replaces the IOSCSIPeripheralType00 driver and everything seems to be going fine. My driver outmatches the generic driver and shows up in the IORegistry just fine. > > However when I use IOLog(...) I don't see my message in /var/log/system.log, instead I am seeing it in the system message buffer. I'm viewing it using "sudo dmesg". Am I missing something? All my references mentioned it should be outputted to /var/log/system.log. Check /var/log/kernel.log . I believe this is where the dmesg content ends up in 10.6 . I hope this helps, Eric > > For reference I'm using OS X 10.6 (i386) and XCode 3.2.1. > > Thanks in advance! > -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/eric.gouriou%40pobox.com > > This email sent to eric.gouriou(a)pobox.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

IOLog printing to system message buffer instead of /var/log/system.log?
by Danny Mardini 06 Jan '10

06 Jan '10

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=ni2+mfRFvWbjUWh58NUJpZ6J/jxrcwEiMmSJxZEa1oI=; b=lgzl4WdGF78M1l6SvX5Ez9Vw003E3Q0OqEzKyQZSHaRFIZwu/52yIndR8ymIMO4ZzG 3VATgumbajdFxz3JwXQaCjWw1uDb7OIKBE8esfBK8TaF47dQ5hn1WqoGfegpjVbpesCm djmSD7DvwmFoOjrCQdf4MY3QkcKgZijJ3qSBo= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:date:message-id:subject:from:to:content-type; b=pbmzvrMjZlNGHBmwVwOyiG5ELtz0MoG+9tdOMq9iGhkedxtG+nWOuyMePisHHjgpfo 7F4kaxyW9CKnKXE9ww0eTBGmmR7OKFX3+epdjpVTQC6aQzojoED5J5/v4cQZ+3fgJQ6L 5SqAlnOAyl28gjH42uPAzu6Vx1RiUBgiSV+JE= _______________________________________________ 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