June 2006 - Darwin-dev - lists.apple.com

Re: GCC fubaring register
by Andrew Pinski 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jun 12, 2006, at 1:32 PM, Ryan Govostes wrote: -- Pinski _______________________________________________ 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… Is there any other way I can tell gcc that I *really* don't want it to use r2 for anything else? The other thing to try is -ffixed-r2. But I don't understand what you are trying to do anyways. This email sent to site_archiver(a)lists.apple.com

1 0

Re: GCC fubaring register
by Andrew Pinski 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jun 12, 2006, at 1:32 PM, Ryan Govostes wrote: What are you really trying to do? Note: asm("mr %0, r2\n\t" : "=r" (toc) : : "r2" ); is wrong as you are not clobbering r2 at all. I'm working with the following function, trimmed down to demonstrate my problem PPC Darwin's ABI does not have TOC and R2 is considered volatile through function calls just like r0. -- Pinski _______________________________________________ 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 fubaring register
by Ryan Govostes 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; b=HrLL1W6s++5YYbyI/FADFkLYsZiW3WLnCuZk6U9qmaTAiOFs1Gk+/qR7k8gdmRZa9319qk26fdt72ZLFYvhg9fIQpNF9TZcbSMpg9HutPU5mWriq4Wc/Ztsi5FZ6JPk+yQD9Wu+SQEHmZUyCRvKG7XiCBVRJmtbW+oyZ4aKKMAc= static void exampleFunc (long *a) { unsigned long toc; asm("mr %0, r2\n\t" : "=r" (toc) : : "r2" ); // ... asm("mr r2, %0\n\t" : : "r" (toc) : ); secondFunc(a); } 0x000020ac mflr r0 0x000020b0 stmw r30,-8(r1) 0x000020b4 stw r0,8(r1) 0x000020b8 stwu r1,-96(r1) 0x000020bc mr r30,r1 0x000020c0 bcl- 20,4*cr7+so,0x20c4 24> 0x000020c4 mflr r31 0x000020c8 stw r3,120(r30) 0x000020cc mr r0,r2 // First inline ASM ... move r2 into r0 0x000020d0 stw r0,56(r30) // store it in memory (unsigned long toc) // Rest of function would go here Thanks, Ryan Govostes _______________________________________________ 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'm working with the following function, trimmed down to demonstrate my problem Essentially, I need to preserve the contents of the TOC (r2) - if it has changed by the time it calls secondFunc(), a horrible crash will ensue. What I've tried to do is to store the contents of the register before doing anything else, and then reset the value immediately before calling secondFunc(). For added protection, the first inline ASM line tells GCC that r2 has been "clobbered," which I understand to mean that it shouldn't be used again throughout the function for any purpose. Looking at the assembly produced by GCC in compiling the above, however, shows that all of this is being disregarded. At 0x000020d0 we're saving r2 to our toc variable, and at 0x000020d8 we're loading it back, but at 0x000020dc GCC has slipped some instructions to overwrite the value of r2, between my inline ASM and the call to secondFunc(). 0x000020d4 lwz r0,56(r30) // Second inline ASM... load toc into r0 0x000020d8 mr r2,r0 // move r0 into r2 -- now we've got the same value we started with 0x000020dc addis r2,r31,0 // But wait, now it's overwriting r2 with something else! 0x000020e0 addi r2,r2,12100 0x000020e4 lwz r0,0(r2) 0x000020e8 lwz r3,120(r30) 0x000020ec mr r12,r0 0x000020f0 mtctr r12 0x000020f4 bctrl // Call to secondFunc() 0x000020f8 lwz r1,0(r1) 0x000020fc lwz r0,8(r1) 0x00002100 mtlr r0 0x00002104 lmw r30,-8(r1) 0x00002108 blr Is there any other way I can tell gcc that I *really* don't want it to use r2 for anything else? I also considered an intermediary function, written entirely in assembly that would restore the TOC and then call the function, but I don't see that as being very graceful. Any thoughts? This email sent to site_archiver(a)lists.apple.com

1 0

dyld binding problem
by Nick Forte 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=beta; d=gmail.com; h=received:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; b=HQ5JDzL7GzW6UUXYxmohTNc3t1+hsYhlaU1DFoDE9A0dObQm88tE69qjLYrlRsN1QQ9NZDboTD3VzMgGDXjegi2sZIqent1CZQGTYA7bgSOH2r9Ec5LxYc6oiQpppfjzl7HYSpAFvMBKhLYX4b+OfvrHkORgv0jk7U5G4cu5lhQ= Thanks, Nick _______________________________________________ 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… Hi, I'm not sure if this is the right list but I'm hoping someone could shed some light on this problem. I have a plugin that dynamically loads a customized version of curl using dlopen(). The application is compiled with -lcurl but my plugin needs to load its own customized version of curl that was built with some changes. The problem is that the application is linking (via -lcurl) with the version of curl that's in /usr/lib/curl (which I don't want to change). As soon as a I do the following: void * dll = dlopen("libcurlNick.dylib", RTLD_LAZY | RTLD_LOCAL ); I see some of the wrong symbols get bound -- some are using my version and some are the version in /usr/lib: dyld: loaded: /usr/lib/libcurl.3.dylib, cpu-sub-type: 3ype: 3 dyld: loaded: /usr/lib/libz.1.dylib, cpu-sub-type: 3 dyld: loaded: /usr/lib/system/libmathCommon.A.dylib, cpu-sub-type: 3 ....... dyld: loaded: /usr/local/lib/libcurlNick.dylib dyld: bind: libcurlNick.dylib:0x0062d060 = libSystem.B.dylib:_malloc, *0x0062d060 = 0x900030fa dyld: bind: libcurlNick.dylib:0x0062d05c = libSystem.B.dylib:_free, *0x0062d05c = 0x90005ae0 dyld: bind: libcurlNick.dylib:0x0062d058 = libSystem.B.dylib:_realloc, *0x0062d058 = 0x9001409f dyld: bind: libcurlNick.dylib:0x0062d054 = libSystem.B.dylib:_strdup, *0x0062d054 = 0x90013bf0 dyld: loaded: /usr/local/lib/libcurlNick.dylibSystem.B.dylib:_calloc, *0x0062d050 = 0x9000803f dyld: bind: libcurlNick.dylib:0x0062d060 = libSystem.B.dylib:_malloc, *0x0062d060 = 0x900030fa000 = 0xa4d87250 dyld: bind: libcurlNick.dylib:0x0062d05c = libSystem.B.dylib:_free, *0x0062d05c = 0x90005ae0x0062e004 = 0xa4d87050 dyld: bind: libcurlNick.dylib:0x0062d058 = libSystem.B.dylib:_realloc, *0x0062d058 = 0x9001409f08 = 0xa4d87220 dyld: bind: libcurlNick.dylib:0x0062d054 = libSystem.B.dylib:_strdup, *0x0062d054 = 0x90013bf000c = 0xa4d8705c dyld: bind: libcurlNick.dylib:0x0062d050 = libSystem.B.dylib:_calloc, *0x0062d050 = 0x9000803f062e010 = 0xa4d87060 ....... dyld: bind: libcurlNick.dylib:_Curl_telnet_done$non_lazy_ptr = libcurl.3.dylib:_Curl_telnet_done, *0x0042e03c = 0x94d9de20 dyld: bind: libcurlNick.dylib:_Curl_tftp_done$non_lazy_ptr = libcurlNick.dylib:_Curl_tftp_done, *0x0042e040 = 0x00427b3c dyld: bind: libcurlNick.dylib:_Curl_http_connect$non_lazy_ptr = libcurl.3.dylib:_Curl_http_connect, *0x0042e044 = 0x94d8ecd3 dyld: bind: libcurlNick.dylib:_Curl_ftp_nextconnect$non_lazy_ptr = libcurl.3.dylib:_Curl_ftp_nextconnect, *0x0042e048 = 0x94d933c7 dyld: bind: libcurlNick.dylib:_Curl_tftp$non_lazy_ptr = libcurlNick.dylib:_Curl_tftp, *0x0042e04c = 0x0042826c dyld: bind: libcurlNick.dylib:_Curl_ftp_done$non_lazy_ptr = libcurl. 3.dylib:_Curl_ftp_done, *0x0042e050 = 0x94d91156 dyld: bind: libcurlNick.dylib:_Curl_ldap$non_lazy_ptr = libcurl. 3.dylib:_Curl_ldap, *0x0042e054 = 0x94d990bc dyld: bind: libcurlNick.dylib:_curl_jmpenv$non_lazy_ptr = libcurl. 3.dylib:_curl_jmpenv, *0x0042e058 = 0xa4d87580 First, why isn't dyld using all the symbols from libcurlNick.dylib -- I promise to you they are there. I want the main application to use / usr/lib/libcurl.3.dylib but I want my plugin inside the application to use /usr/local/lib/libcurlNick.dylib. What am I doing wrong -- is this even possible to do? This email sent to site_archiver(a)lists.apple.com

1 0

Re: DiskArb DiskChangedCallback bug?
by Brian Bergstrand 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Jun 12, 2006, at 1:40 PM, Dan Markarian wrote: Hi Brian, Thanks. P.S. It would be nice to see the documentation clarify this situation. Dan Le 12 Jun 2006 à 1:58 PM, Brian Bergstrand a écrit : -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Jun 12, 2006, at 12:49 PM, Chris Sarcone wrote: Brian -- Nevermind. I was looking at the wrong API. extern CFArrayRef kDADiskDescriptionWatchVolumePath; Thanks for the quick response! -- Chris On Jun 12, 2006, at 10:42 AM, Chris Sarcone wrote: Brian -- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin) -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin) iD8DBQFEjcBOedHYW7bHtqIRAsbWAJ9bA5vk1T33RTXI/FoGuX3QRoag0gCcDcBy f+552tZHJd99hZUG3spLkMA= =nFKb -----END PGP SIGNATURE----- _______________________________________________ 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… DADiskDescriptionChangedCallback does notify you when a property of a disk has changed, including whether it is mounted or not, however, it does not notify you of the existence of the disk. You must also subscribe to DADiskAppearedCallback and DADiskDisappearedCallback. It is possible that a disk is already mounted when we tell you of its existence. Ok, my thinking was, the disk is inserted after I've already registered my callback so a change from attached to mounted would be sent. When a disk is attached, there is no mount point, but apparently DA can see attach/mount as a single event and that's where the confusion was. I've switched to DADiskAppearedCallback to detect a mount and still use Changed to detect the unmount. This now works. As for using DARegisterDiskDescriptionChangedCallback() for mounts, I think that's incorrect. DARegisterDiskDescriptionChangedCallback() is meant to track disk that are already mounted and changing either their name or path. I think your really want DARegisterDiskAppearedCallback(). Hmm, won't that be called when the disk appears in IOKit and not when it's mounted? I don't really care about the former. Or will it be called both for attach and mount? If that's the case, then I could use the volumePath key in the Disk dictionary to determine if it's being mounted. Also, if ChangedCallback() is not supposed to be used, then the documentation should be clarified: /*! * @constant kDADiskDescriptionWatchVolumePath * Predefined CFArray object containing a set of disk description keys appropriate for * watching volume mount changes using DARegisterDiskDescriptionChangedCallback(). */ I take "watching volume mount changes", to mean changing from mounted - -> unmounted OR vise-versa. "watching volume un-mounts" would be a better term if that is all ChangedCallback does. DARegisterDiskDescriptionChangedCallback(daSession, NULL, kDADiskDescriptionWatchVolumePath, DiskArbCallback_ChangeNotification, NULL); The third argument is supposed to be a CFArrayRef of keys to watch, not a CFDictionaryRef. You might want to try fixing that, then retest... iD8DBQFEjatIedHYW7bHtqIRAjrnAKDueVlFzcor+NdyG0haXVfsaWMVjwCcCqbv hDuY/92TfgL7KQWq9/yEUqo= =yBfx -----END PGP SIGNATURE----- _______________________________________________ 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/markarian% 40apple.com This email sent to markarian(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: DiskArb DiskChangedCallback bug?
by Dan Markarian 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Brian, Dan Le 12 Jun 2006 à 1:58 PM, Brian Bergstrand a écrit : -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Jun 12, 2006, at 12:49 PM, Chris Sarcone wrote: Brian -- Nevermind. I was looking at the wrong API. extern CFArrayRef kDADiskDescriptionWatchVolumePath; Thanks for the quick response! -- Chris On Jun 12, 2006, at 10:42 AM, Chris Sarcone wrote: Brian -- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin) _______________________________________________ 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… DADiskDescriptionChangedCallback does notify you when a property of a disk has changed, including whether it is mounted or not, however, it does not notify you of the existence of the disk. You must also subscribe to DADiskAppearedCallback and DADiskDisappearedCallback. It is possible that a disk is already mounted when we tell you of its existence. As for using DARegisterDiskDescriptionChangedCallback() for mounts, I think that's incorrect. DARegisterDiskDescriptionChangedCallback() is meant to track disk that are already mounted and changing either their name or path. I think your really want DARegisterDiskAppearedCallback(). Hmm, won't that be called when the disk appears in IOKit and not when it's mounted? I don't really care about the former. Or will it be called both for attach and mount? If that's the case, then I could use the volumePath key in the Disk dictionary to determine if it's being mounted. Also, if ChangedCallback() is not supposed to be used, then the documentation should be clarified: /*! * @constant kDADiskDescriptionWatchVolumePath * Predefined CFArray object containing a set of disk description keys appropriate for * watching volume mount changes using DARegisterDiskDescriptionChangedCallback(). */ I take "watching volume mount changes", to mean changing from mounted - -> unmounted OR vise-versa. "watching volume un-mounts" would be a better term if that is all ChangedCallback does. DARegisterDiskDescriptionChangedCallback(daSession, NULL, kDADiskDescriptionWatchVolumePath, DiskArbCallback_ChangeNotification, NULL); The third argument is supposed to be a CFArrayRef of keys to watch, not a CFDictionaryRef. You might want to try fixing that, then retest... iD8DBQFEjatIedHYW7bHtqIRAjrnAKDueVlFzcor+NdyG0haXVfsaWMVjwCcCqbv hDuY/92TfgL7KQWq9/yEUqo= =yBfx -----END PGP SIGNATURE----- _______________________________________________ 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/markarian% 40apple.com This email sent to markarian(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: kqueue - supported filesystems?
by Dave Zarzycki 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jun 12, 2006, at 10:55 AM, Josh Graessley wrote: On Jun 12, 2006, at 8:34 AM, Dave Zarzycki wrote: On Jun 9, 2006, at 6:31 PM, Jim Correia wrote: The kqueue man page says: davez _______________________________________________ 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… Not all filesystem types support kqueue-style notifications. And even some that do, like some remote filesystems, may only support a subset of the notification semantics described here. Which filesystems are supported? If memory serves me correct, just HFS in Tiger. But we're trying (no promises) to move that logic up to the VFS layer so that all file systems get it for free. It is worth noting that even once this is at the VFS layer, there will still be issues. I believe that network file systems are unlikely to support this. The protocol would need to push this information to the clients. Well, given a VFS layer implementation, the notifications would be local only, much like memory mapping a file on a network volume can only, at best, share memory with local processes. This email sent to site_archiver(a)lists.apple.com

1 0

Re: DiskArb DiskChangedCallback bug?
by Brian Bergstrand 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Jun 12, 2006, at 12:49 PM, Chris Sarcone wrote: Brian -- Nevermind. I was looking at the wrong API. extern CFArrayRef kDADiskDescriptionWatchVolumePath; Thanks for the quick response! -- Chris On Jun 12, 2006, at 10:42 AM, Chris Sarcone wrote: Brian -- -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.3 (Darwin) iD8DBQFEjatIedHYW7bHtqIRAjrnAKDueVlFzcor+NdyG0haXVfsaWMVjwCcCqbv hDuY/92TfgL7KQWq9/yEUqo= =yBfx -----END PGP SIGNATURE----- _______________________________________________ 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… As for using DARegisterDiskDescriptionChangedCallback() for mounts, I think that's incorrect. DARegisterDiskDescriptionChangedCallback () is meant to track disk that are already mounted and changing either their name or path. I think your really want DARegisterDiskAppearedCallback(). Hmm, won't that be called when the disk appears in IOKit and not when it's mounted? I don't really care about the former. Or will it be called both for attach and mount? If that's the case, then I could use the volumePath key in the Disk dictionary to determine if it's being mounted. Also, if ChangedCallback() is not supposed to be used, then the documentation should be clarified: /*! * @constant kDADiskDescriptionWatchVolumePath * Predefined CFArray object containing a set of disk description keys appropriate for * watching volume mount changes using DARegisterDiskDescriptionChangedCallback(). */ I take "watching volume mount changes", to mean changing from mounted - -> unmounted OR vise-versa. "watching volume un-mounts" would be a better term if that is all ChangedCallback does. DARegisterDiskDescriptionChangedCallback(daSession, NULL, kDADiskDescriptionWatchVolumePath, DiskArbCallback_ChangeNotification, NULL); The third argument is supposed to be a CFArrayRef of keys to watch, not a CFDictionaryRef. You might want to try fixing that, then retest... This email sent to site_archiver(a)lists.apple.com

1 0

Re: kqueue - supported filesystems?
by Josh Graessley 12 Jun '06

12 Jun '06

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jun 12, 2006, at 8:34 AM, Dave Zarzycki wrote: On Jun 9, 2006, at 6:31 PM, Jim Correia wrote: The kqueue man page says: -josh _______________________________________________ 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 Not all filesystem types support kqueue-style notifications. And even some that do, like some remote filesystems, may only support a subset of the notification semantics described here. Which filesystems are supported? If memory serves me correct, just HFS in Tiger. But we're trying (no promises) to move that logic up to the VFS layer so that all file systems get it for free. It is worth noting that even once this is at the VFS layer, there will still be issues. I believe that network file systems are unlikely to support this. The protocol would need to push this information to the clients. smime.p7s

1 0

Re: DiskArb DiskChangedCallback bug?
by Chris Sarcone 12 Jun '06

12 Jun '06

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