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

Re: gcc4 and casting lvalues
by Andrew Pinski 14 Jan '05

14 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jan 13, 2005, at 7:53 PM, Sam Vaughan wrote: It appears that gcc 4 has an issue with casting lvalues: Is this expected behaviour? I might have a lot of code to change... Yes it is expected, this was a bad extension to C which was removed. You can read the release notes for gcc 4.0.0 at <http://gcc.gnu.org/gcc-4.0/changes.html>. Thanks, Andrew 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

gcc4 and casting lvalues
by Sam Vaughan 14 Jan '05

14 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com It appears that gcc 4 has an issue with casting lvalues: % cat > test.c #include <stdio.h> int main() { void* vp = 0; #ifdef A (long*)vp += 1; #endif #ifdef B vp = (long*)vp + 1; #endif printf("vp: %p\n", vp); return 0; } % gcc-4.0 -DA test.c && ./a.out test.c: In function 'main': test.c:8: error: invalid lvalue in assignment % gcc-4.0 -DB test.c && ./a.out vp: 0x4 % gcc-3.3 -DA test.c && ./a.out vp: 0x4 % gcc-3.3 -DB test.c && ./a.out vp: 0x4 Is this expected behaviour? I might have a lot of code to change... Sam _______________________________________________ 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

Building Apple's gcc-1495 on linux x86 for cross-compiling
by Adam Cohen 13 Jan '05

13 Jan '05

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:message-id:date:from:reply-to:to:subject:mime-version:content-type:content-transfer-encoding; b=gAQ9sa0Pn9OpMnemx6HaQWVhQ1AFgvqIZLh0zM/mTxOIOwiZl5qYRwzZRT/dAk25ZpGy7h6jEFqsnw1bdN3Ftm0i+iPK3DmoDk3X1ZmpTPZ3ltOJamV491e9rbEZQzf/c3OKvweNUMxRygUVE1A9f1s/0Az4Vd3UY1LyO4a4pcU= I've been trying without much success to build Apple's copy of gcc-1495 (or build 1671) on an x86 machine running debian sid, with gcc 3.3.5 (Debian 1:3.3.5-6) in order to use distcc from within Xcode to cross-compile Cocoa projects. I've tried the instructions listed here: <http://www.myownlittleworld.com/computers/darwin-cross-distcc.html> (which uses a patched version of Apple's cctools), but the compile fails with the following error: make[2]: Entering directory `/home/build/crossdev/objdir/libiberty/testsuite' make[2]: Nothing to be done for `install'. make[2]: Leaving directory `/home/build/crossdev/objdir/libiberty/testsuite' make[1]: Leaving directory `/home/build/crossdev/objdir/libiberty' make: Target `install' not remade because of errors. I've also tried the instructions here: <http://www.opendarwin.org/pipermail/odcctools/2004-November/000034.html> (which use odcctools), and the following configure directive: ../configure --prefix=/usr/local/crossdev --target=powerpc-apple-darwin --with-as=/usr/local/crossdev/bin/powerpc-apple-darwin-as --with-ld=/usr/local/crossdev/bin/powerpc-apple-darwin-ld --with-sysroot=/home/adam/darwin-root --enable-static --enable-shared --enable-languages=c,objc but the compile fails with the error: In file included from /usr/include/netinet/in.h:23, from ../../gcc/genindex.c:20: ../../more-hdrs/stdint.h:24:27: machine/types.h: No such file or directory even though my configuration directive specifies --with-sysroot to point to the directory where all my OS X headers have been copied.. (darwin-root contains usr/include, usr/lib and System/Library/Frameworks) If I add "-I/home/adam/darwin-root/usr/include/" to CFLAGS and then recompile, I get: /home/adam/darwin-root/usr/include/libkern/OSByteOrder.h:561:2: #error Unknown endianess. I can get around this by including -D__LITTLE_ENDIAN__ in CFLAGS as well, but then I get: In file included from ../../gcc/intl.h:34, from ../../gcc/c-opts.c:32: /usr/include/libintl.h: At top level: /usr/include/libintl.h:47: error: parse error before "__attribute_format_arg__" /usr/include/libintl.h:47: error: ISO C forbids data definition with no type or storage class So I'm just wondering if anyone on the list has successfully compiled a copy of Apple's gcc-1495 (or 1671) on any Linux distribution, and if so, what version of cctools, gcc, etc, were used. Thanks in advance, Adam P.S. I should mention that I have been able to successfully compile the FSF version of gcc-3.4.3 on x86 using odcctools, although when attempting to cross-compile from Xcode using distcc, I get errors regarding Apple-specific compiler flags, such as: cc1obj: error: unrecognized command line option "-arch" cc1obj: error: unrecognized command line option "-fpascal-strings" cc1obj: error: unrecognized command line option "-ffix-and-continue" cc1obj: error: unrecognized command line option "-fpch-preprocess" _______________________________________________ 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: Disk Usage Incorrect
by Michael Hogsett 13 Jan '05

13 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mozilla Thunderbird 0.7.3 (Windows/20040803) - Mike Hogsett _______________________________________________ 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… Matt Deatherage wrote: If there are other explanations, I can't think of them. Another potential, although unlikely, possibility is that a running process has an open file handle to a file that has been deleted from the filesystem. If MacOSX behaves like other unix OSs the file will continue to occupy space on the disk until the filehandle is closed, but since there is no entry in the file system for this file 'du' will not include the deleted-but-open file's size in its calculation since it can't stat the file. For 58GBytes of missing space this is highly unlikely. Also this is not possible across reboots (since obviously all filehandles are closed on shutdown). This email sent to site_archiver(a)lists.apple.com

1 0

Re: Disk Usage Incorrect
by seebs＠plethora.net 13 Jan '05

13 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com In message <b02010501-1037-8D3666BA65B811D9B75D000D932C95B4(a)[192.168.1.3]>, Mat t Deatherage writes: >Unless I am mistaken, Mac OS X's HFS Plus implementation does not >allocate any blocks to a fork until that fork has a non-zero length. >If you have a copy of Norton Disk Editor X, you can look at the >HFS Plus directory entry for any file and see how many blocks are >allocated to each fork. ... Norton did a disk editor? I recall them discontinuing this in Norton Utilities for Mac many years ago. It's been removed from the Windows stuff too. Is there some hidden way to get to it? -s _______________________________________________ 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: Disk Usage Incorrect
by Matt Deatherage 13 Jan '05

13 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 1/12/05 at 5:17 PM, Craig E. Ward <cew(a)ACM.ORG> wrote: > If a resource fork is also allocated for a 1 byte file, wouldn't that > explain the extra 4 blocks? Unless I am mistaken, Mac OS X's HFS Plus implementation does not allocate any blocks to a fork until that fork has a non-zero length. If you have a copy of Norton Disk Editor X, you can look at the HFS Plus directory entry for any file and see how many blocks are allocated to each fork. On 1/12/05 at 4:34 PM, Jerry Pendergraft <jerry(a)endocardial.com> wrote: > On the point of comparing file size to allocated size I did a quick and > dirty perl script to stat files and find out what allocated size was for > various file sizes. Got a big surprise when this reported more than a > fraction of a block of waste for larger files. And the the minimum > allocation seems to be 8 blocks even for the 1 byte file. On most HFS Plus disks, the allocation block size is 4KB, as noted; since stat uses 512-byte blocks (as also noted by Clark Cox), it will *usually* tell you that there's more than one 512-byte block "wasted" at the end of the file. Presuming a linear distribution of file sizes across a disk, each file wastes, on average, half the allocation block size in its last block, or 2KB. Since you say the disk holds 860,137 files, you could expect to see about 1.7GB of space that's allocated but not used within any file's logical EOF. You report 58GB missing, and that's way too much. In fact, it's outside the possible bounds for a normal HFS Plus disk. The most any file can waste is the size of a single allocation block minus one byte, and on a standard HFS Plus disk, that's 4KB-1 byte, or 4095 bytes. Multiply that times 860,137 files and you come up with about 3.5GB of "wasted" space - only 6% the figure you report. So what could be happening? You say the system's disk repair framework (used in both fsck and Disk Utility) says nothing's wrong with the disk, so that's out. You might have a volume with larger allocation blocks, but that's easy enough to determine: echo "" > testfile.txt and look at it in the Finder with "Get Info." The "(1 byte)" is the logical EOF, the other size number is the size of one allocation block on your disk. I'll bet it's 4KB, though. (It's possible to create HFS Plus disks with allocation blocks as small as 512 bytes, but Mac OS X's file system code is optimized for 4KB blocks, so it's generally not worth doing.) If this is a test system that crashes a lot, you could have a lot of hidden wasted space in your test files. HFS Plus allocates by clumps for efficiency and to avoid fragmentation - opening a file and writing one byte to it allocates, by default, 16 allocation blocks (64KB) to the file. However, *closing* the file releases all the completely-unused allocation blocks back to the system. Terminating a process should do the same thing, but I can't vouch for it. The Finder knows both the "physical size" (allocation blocks) and "size" (logical EOF) of every item, so in theory, you can write an AppleScript to find all files whose physical size - size > 4095 bytes to see what's going on. (Unfortunately, either of those can be "missing value" on folders or packages, and Finder doesn't really understand the concept of a packaged document, like a GarageBand song, so my quick attempts to do it all choked on such packages in ways I can't quickly work around.) Nonetheless, if you can do it in AppleScript, I imagine you can do it in shell scripting too. I suspect that an earlier post is right too, though - utilities not run in admin mode may not be able to properly count files and folders to which you have no read access. You may have a lot more files than you think. You can always try launching a tool as root to see if it finds more reasonable numbers. If the disk is not damaged, and you're missing 58GB of space, then either you can't see the files occupying it, or lots of files have space allocated beyond the last allocation block required to hold the logical EOF. If there are other explanations, I can't think of them. --Matt -- Matt Deatherage <mattd(a)macjournals.com> <http://www.macjournals.com> I read this list in digest mode; copy me privately for faster responses _______________________________________________ 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: Disk Usage Incorrect [solved]
by Shawn Erickson 13 Jan '05

13 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jan 13, 2005, at 12:46 AM, Jonas Maebe wrote: On 12 jan 2005, at 23:47, Eric Ocean wrote: -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… FWIW, OmniDiskSweeper reports /Users/anthem to be ~850MB when du says it's really ~23GB. Other user folders have similar discrepancies (though not as large). The problem is not caused by lots of files; the space in question is occupied by music, mostly, that was dropped in the ~/Public/Drop Box folder. I guess that warrants a bug report to the Omni people... Well it is limited by file access permission like every other application. So unless it is being run with admin privileges it will not be able to see all files and directories. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Disk Usage Incorrect [solved]
by Jonas Maebe 13 Jan '05

13 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 12 jan 2005, at 23:47, Eric Ocean wrote: I guess that warrants a bug report to the Omni people... Jonas _______________________________________________ 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… FWIW, OmniDiskSweeper reports /Users/anthem to be ~850MB when du says it's really ~23GB. Other user folders have similar discrepancies (though not as large). The problem is not caused by lots of files; the space in question is occupied by music, mostly, that was dropped in the ~/Public/Drop Box folder. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Disk Usage Incorrect
by Clark Cox 12 Jan '05

12 Jan '05

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:message-id:date:from:reply-to:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:references; b=WYsLEfISuJvu7PXnZXgwo7Sz5gSoNaNRl828h+44KiKMK/AJp/lRxBibOdoHyqVurIyWhayjVkq+9WcsJMWFsolmbC1xHqppbqSH6bLqPKjHPI+02BQFv3sjNQQykBm4Z4TnoiFLv/dhm62/ITx4OWkWWvx3d6czOiRE/SWXoi8= On Wed, 12 Jan 2005 16:34:07 -0600 (CST), Jerry Pendergraft <jerry(a)endocardial.com> wrote: > On the point of comparing file size to allocated size I did a quick and > dirty perl script to stat files and find out what allocated size was for > various file sizes. Got a big surprise when this reported more than a > fraction of a block of waste for larger files. And the the minimum > allocation seems to be 8 blocks even for the 1 byte file. > > Does darwin set the stat struct variable st_blocks somehow different than > the man page describes as /* blockes allocated for file */ ?? In the man page, it states that stat reports the number of 512B blocks, while the minimum filesize is 4K ... so nothing fishy here (4K = 512B * 8) -- Clark S. Cox III clarkcox3(a)gmail.com http://www.livejournal.com/users/clarkcox3/ http://homepage.mac.com/clarkcox3/ _______________________________________________ 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: Disk Usage Incorrect
by Craig E. Ward 12 Jan '05

12 Jan '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com ... fraction of a block of waste for larger files. And the the minimum allocation seems to be 8 blocks even for the 1 byte file. At 4:34 PM -0600 1/12/05, Jerry Pendergraft wrote: If a resource fork is also allocated for a 1 byte file, wouldn't that explain the extra 4 blocks? -- Internet: cew(a)ACM.ORG "There are 10 kinds of people in the world: Those who understand binary and those who don't." _______________________________________________ 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