November 2007 - Darwin-kernel

Re: pthread_cancel and cancelation points still broken in Mac OS X 10.5 Leopard?
by Michael Smith 22 Nov '07

22 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Per, = Mike On Nov 22, 2007, at 2:41 AM, Per Mildner wrote: The Single Unix Specification v3 (SUSv3) defines cancelation points in http://www.opengroup.org/onlinepubs/000095399/functions/xsh_chap02_09.html#… int read(...) { pthread_testcancel(); return read_nocancel(...); } Transcript: bash$ gcc cancel_bug.c -Wall && ./a.out Created thread, sleeping calling read..<press return> .called read()==1 calling read..<press return> .called read()==1 calling read..cancelling read() <blocks forever or until more input is available> /* cancel_bug.c */ #include <assert.h> #include <pthread.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> static int exit_status = EXIT_FAILURE; void * func(void *arg) { char buf; int oldstate; ssize_t res; return NULL; } int main(void) { pthread_t thread; void *retval; CHECK(pthread_create(&thread, NULL, func, NULL) == 0); fprintf(stderr, "Created thread, sleeping\n");fflush(stderr); sleep(10); /* ensure thread reaches blocking read */ fprintf(stderr, "cancelling read()\n");fflush(stderr); CHECK(pthread_cancel(thread) == 0); CHECK(pthread_join(thread, &retval) == 0); CHECK(PTHREAD_CANCELED == retval); return exit_status; } Regards, _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/drivers%40mu.org _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Whilst your complaint may well be legitimate, this is not a bug- reporting forum. Please file a bug (bugreporter.apple.com) and include the bug number when discussing the matter here or (ideally) in any other forum. My tests indicate that pthread_cancel still cannot interrupt blocking system calls like read(2) in Intel Mac OS X 10.5.1, despite the claim that it is now UNIX 2003 compliant. A brief look at the Darwin 9.0 kernel sources seems to confirm this. The critical part, and the reason why pthread_testcancel is not sufficient is "If a thread has cancelability enabled and a cancellation request is made with the thread as a target while the thread is suspended at a cancellation point, the thread shall be awakened and the cancellation request shall be acted upon.". That is, pthread_cancel should wake (a well defined list of) blocking system calls. The below test program shows that a blocking read() is _not_ awakened by pthread_cancel, making the cancelation handling in Leopard no improvement over earlier OS releases and in violation of the SUSv3 standard. The test program starts a worker thread that loops reading from standard input. The main thread then sleeps for a while to ensure that the worker thread has blocked in read(2) and the main thread then cancels the worker thread. This program terminates on Linux and Solaris but not on Intel Leopard 10.5.1. >From a cursory glance at xnu-1228, e.g. sys_generic.c, it looks as if the system calls that SUSv3 requires to be cancelation points are all implemented something like the following template: That is, if a cancel is pending when the system call is entered then the call will be canceled but once the real system call is blocking it will not react to cancel requests. This is with Xcode 3.0. Compiling with -D_APPLE_C_SOURCE does not make a difference. #define CHECK(X) do { if ((X) == 0) { fprintf(stderr, "%s:%d CHECK FAILED\n", __FILE__, (int)__LINE__); } } while (0) exit_status = EXIT_SUCCESS; /* redundant */ CHECK(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &oldstate) == 0); /* Will not terminate unless cancelled or read error */ do { fprintf(stderr, "calling read..");fflush(stderr); res = read(STDIN_FILENO, &buf, 1); fprintf(stderr, ".called read()==%ld\n", (long)res);fflush(stderr); } while (res != -1); So, has Apple and the Open Group conformance testers made a huge mistake or am I missing something? This email sent to drivers(a)mu.org This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Rick Macklem 22 Nov '07

22 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Just a clarification. vn_rdwr() is not part of a KPI and should only be used for a research project, or similar, imho. I have an nfsv4 client that does a bunch of stuff beyond the KPIs, but I consider it an open source pseudo-research (maybe some useful performance results someday) project. For a chuckle, you could grab the sources anonymous ftp from: ftp.cis.uoguelph.ca/pub/nfsv4/darwin-port/taiger-client.tar.gz and grep for "Holy Bogosity Batman" in src/newnfs/newnfs.c. Good luck with writing from the kernel, rick _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

pthread_cancel and cancelation points still broken in Mac OS X 10.5 Leopard?
by Per Mildner 22 Nov '07

22 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com The Single Unix Specification v3 (SUSv3) defines cancelation points in http://www.opengroup.org/onlinepubs/000095399/functions/xsh_chap02_09.html#… int read(...) { pthread_testcancel(); return read_nocancel(...); } Transcript: bash$ gcc cancel_bug.c -Wall && ./a.out Created thread, sleeping calling read..<press return> .called read()==1 calling read..<press return> .called read()==1 calling read..cancelling read() <blocks forever or until more input is available> /* cancel_bug.c */ #include <assert.h> #include <pthread.h> #include <stdio.h> #include <stdlib.h> #include <unistd.h> static int exit_status = EXIT_FAILURE; void * func(void *arg) { char buf; int oldstate; ssize_t res; return NULL; } int main(void) { pthread_t thread; void *retval; CHECK(pthread_create(&thread, NULL, func, NULL) == 0); fprintf(stderr, "Created thread, sleeping\n");fflush(stderr); sleep(10); /* ensure thread reaches blocking read */ fprintf(stderr, "cancelling read()\n");fflush(stderr); CHECK(pthread_cancel(thread) == 0); CHECK(pthread_join(thread, &retval) == 0); CHECK(PTHREAD_CANCELED == retval); return exit_status; } Regards, _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… My tests indicate that pthread_cancel still cannot interrupt blocking system calls like read(2) in Intel Mac OS X 10.5.1, despite the claim that it is now UNIX 2003 compliant. A brief look at the Darwin 9.0 kernel sources seems to confirm this. The critical part, and the reason why pthread_testcancel is not sufficient is "If a thread has cancelability enabled and a cancellation request is made with the thread as a target while the thread is suspended at a cancellation point, the thread shall be awakened and the cancellation request shall be acted upon.". That is, pthread_cancel should wake (a well defined list of) blocking system calls. The below test program shows that a blocking read() is _not_ awakened by pthread_cancel, making the cancelation handling in Leopard no improvement over earlier OS releases and in violation of the SUSv3 standard. The test program starts a worker thread that loops reading from standard input. The main thread then sleeps for a while to ensure that the worker thread has blocked in read(2) and the main thread then cancels the worker thread. This program terminates on Linux and Solaris but not on Intel Leopard 10.5.1. >From a cursory glance at xnu-1228, e.g. sys_generic.c, it looks as if the system calls that SUSv3 requires to be cancelation points are all implemented something like the following template: That is, if a cancel is pending when the system call is entered then the call will be canceled but once the real system call is blocking it will not react to cancel requests. This is with Xcode 3.0. Compiling with -D_APPLE_C_SOURCE does not make a difference. #define CHECK(X) do { if ((X) == 0) { fprintf(stderr, "%s:%d CHECK FAILED\n", __FILE__, (int)__LINE__); } } while (0) exit_status = EXIT_SUCCESS; /* redundant */ CHECK(pthread_setcancelstate(PTHREAD_CANCEL_ENABLE, &oldstate) == 0); /* Will not terminate unless cancelled or read error */ do { fprintf(stderr, "calling read..");fflush(stderr); res = read(STDIN_FILENO, &buf, 1); fprintf(stderr, ".called read()==%ld\n", (long)res);fflush(stderr); } while (res != -1); So, has Apple and the Open Group conformance testers made a huge mistake or am I missing something? This email sent to site_archiver(a)lists.apple.com

1 0

missing IOLog entries in /var/log/system.log, redux
by Chris Neustrup 21 Nov '07

21 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com User-agent: Thunderbird 1.5.0.7 (X11/20070925) More detail about the lost IOLog entries. It appears that the entries are being made, but they fall off the end of the file, or something similar. If I add too many, or perhaps they arrive in the file too often, then they go to the bit bucket. So it looks like the system.log file cannot be appended to over a certain capacity or speed. When I fill it with lots of lines, they get severely truncated. Now I make sure that I'm not sending too many, and I can see all of them. It's kind of a pain to have to clear out IOLog calls from the source to get more of them. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: MTRRs and PATs
by Derek Kumar 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Victor, For example, on a Mac Pro running Leopard: This is the MTRR capabilities MSR (0xFE): reggie_se -r -a 0xfe -B 64 (gdb) p mtrr_state $5 = { MTRRcap = 0x508, MTRRdefType = 0xc00, var_range = 0x6483e00, var_count = 0x8, fix_range = {{ types = 0x606060606060606 }, { types = 0x606060606060606 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }, { types = 0x0 }} } Derek On Nov 20, 2007, at 3:34 PM, Victor Vedovato wrote: Thanks, Victor _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/drk%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I presume this is for debugging purposes? You can use the reggie_se tool from the CHUD toolkit to display the contents of various MSRs by number. reggie_se 4.5.0 CPU0 - 3000MHz Intel Xeon 53XX Address 0x000000fe 0x0000000000000508 |0 0 0 0|0 0 0 0|0 0 0 0| 0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 0 0 0|0 1 0 1|0 0 0 0|1 0 0 0| i.e. there are 8 variable ranges, WC and fixed range registers are supported The contents of IA32_CR_PAT (0x277): bash-3.2# reggie_se -r -a 0x277 -B 64 reggie_se 4.5.0 CPU0 - 3000MHz Intel Xeon 53XX Address 0x00000277 0x0001040600070406 |0 0 0 0|0 0 0 0|0 0 0 0| 0 0 0 1|0 0 0 0|0 1 0 0|0 0 0 0|0 1 1 0|0 0 0 0|0 0 0 0|0 0 0 0|0 1 1 1|0 0 0 0|0 1 0 0|0 0 0 0|0 1 1 0| i.e. the PA6 field (which is what the kernel's MMU management (pmap) layer selects when asked to mark mappings WC) is set to write combine. In general, the PAT and page table based mechanisms are used preferentially by the kernel's pmap layer (and indirectly by the IOKit APIs). This is a dump of the kernel's internal MTRR state structure (obtained via the live kernel debugger). The var_range field is an array containing base and mask values (specifying base, cache attribute type and range) that can be introspected. Standard disclaimer: manipulating these MSRs directly can lead to unpredictable results; they should be altered directly by the kernel alone. You must, of course, use the provided APIs (such as IOSetProcessorCacheMode) to set cache attributes. Anyone know of a tool for examining x86 MTRR and PAT registers or displaying memory range cache controls on OS X? This email sent to drk(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Terry Lambert 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -- Terry On Nov 20, 2007, at 2:58 PM, Terry Lambert <tlambert(a)apple.com> wrote: #define path "mypath" and then #includes a system header to get e.g.: Int open(const char *path, int flags, ...); -- Terry On Nov 20, 2007, at 8:01 AM, Quinn <eskimo1(a)apple.com> wrote: [1] The bug tracking this is <rdar://problem/3524590>. [2] Alas, there are fairly good technical reasons for this. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/tlambert%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/tlambert%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Of course, that's supposed to be "prototype parameters either", but my iPhone ate my homework. 8^). Actually both UNIX/POSIX and ANSI/ISO C standards require that prototypeseither be omitted or in the implementation name space -- have underscore prefixes so if a user e.g. does: In scope, path in the parameter list doesn't change into something else that's syntactically incorrect (or worse, syntactically correct but wrong). Unfortunately, there's user code out there that puts variables and definitions into that space anyway, so we just try to avoid it all. At 9:54 -0600 20/11/07, Kynan Shook wrote: Now, is there any documentation on these interfaces? A google on vnode_open returns very few results... Not really )-: [1] And given the BSD habit of not including formal parameter names in prototypes [2], your only real option is to look at the Darwin source. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/ > Apple Developer Relations, Developer Technical Support, Core OS/ Hardware This email sent to tlambert(a)apple.com This email sent to tlambert(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Terry Lambert 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com #define path "mypath" and then #includes a system header to get e.g.: Int open(const char *path, int flags, ...); -- Terry On Nov 20, 2007, at 8:01 AM, Quinn <eskimo1(a)apple.com> wrote: [1] The bug tracking this is <rdar://problem/3524590>. [2] Alas, there are fairly good technical reasons for this. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/tlambert%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Actually both UNIX/POSIX and ANSI/ISO C standards require that prototypeseither be omitted or in the implementation name space -- have underscore prefixes so if a user e.g. does: In scope, path in the parameter list doesn't change into something else that's syntactically incorrect (or worse, syntactically correct but wrong). Unfortunately, there's user code out there that puts variables and definitions into that space anyway, so we just try to avoid it all. At 9:54 -0600 20/11/07, Kynan Shook wrote: Now, is there any documentation on these interfaces? A google on vnode_open returns very few results... Not really )-: [1] And given the BSD habit of not including formal parameter names in prototypes [2], your only real option is to look at the Darwin source. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/ > Apple Developer Relations, Developer Technical Support, Core OS/ Hardware This email sent to tlambert(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Terry Lambert 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com -- Terry On Nov 20, 2007, at 7:54 AM, Kynan Shook <kshook(a)cae.wisc.edu> wrote: Thanks for the gun, I promise I won't shoot myself in the foot. ;-) On Nov 20, 2007, at 5:05 AM, Quinn wrote: In that case I would just write it to a file (-: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/tlambert%40apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Internal documentation in source files is pretty significantly improved in Leopard. It's not everywhere, but we had to do this for errno origin and value tracking for the UNIX conformance work. Mostly this is under xnu/bsd/, but that's where you look to be looking anyway. I'd definitely agree with Quinn that if this were anything but a research project... well. Now, is there any documentation on these interfaces? A google on vnode_open returns very few results... Kynan It's not that you /can't/ read or write files from the kernel, it's that doing this exposes you to all sorts of crazy deadlock situations. In this case, you're sitting at the block driver level and you end up calling all the way back up to the VFS level. It's very easy to deadlock due to reentrancy. It's also easy to deadlock due to lack of resources. OTOH, for a research project you should be able to make it work. You can get the vnode for a path using vnode_lookup. You can open it using vnode_open. And you can write to it using VNOP_WRITE. You don't even have to take a dependency on the kernel; these are all KPIs. This email sent to tlambert(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

MTRRs and PATs
by Victor Vedovato 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thanks, Victor _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… Anyone know of a tool for examining x86 MTRR and PAT registers or displaying memory range cache controls on OS X? This email sent to site_archiver(a)lists.apple.com

1 0

Re: Writing from the kernel
by Rick Macklem 20 Nov '07

20 Nov '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Tue, 20 Nov 2007, Quinn wrote: In that case I would just write it to a file (-: Although I haven't done it in Darwin since 10.3, I find vn_rdwr() pretty handy for doing this. I find it's easier to open the file in userland and pass the file descriptor into the kernel than do the open, etc. from inside the kernel. You can easily get a vnode pointer from the file desciptor. As Quinn said, xnu-792 (or maybe xnu-1226 now) sources are the doc. grep on vn_rdwr() in xnu-792.x.y/bsd/kern should give you some hints. rick _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… At 1:30 -0600 20/11/07, Kynan Shook wrote: So, what is a good way to get the data out of the kernel? I'm looking for fast and simple here - it's only a project, and will only ever be used on my machine, so minimizing the time spent on implementation is the priority. [good stuff snipped] OTOH, for a research project you should be able to make it work. You can get the vnode for a path using vnode_lookup. You can open it using vnode_open. And you can write to it using VNOP_WRITE. You don't even have to take a dependency on the kernel; these are all KPIs. This email sent to site_archiver(a)lists.apple.com

1 0