May 2005 - Darwin-kernel - lists.apple.com

Read through Filter Scheme Driver
by Imran Kazi 10 May '05

10 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcVVZvm/u5qeCyZbRmaygmNPsV0otQ== _______________________________________________ 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: Write on specific IOMedia through filter driver
by Imran Kazi 10 May '05

10 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcVSgsXwBsWwqdgNRYS4sZ5dv7tFkAC32TAw Dear Dan As per your previous mail, we have change the partition name of the volumes, and use media->getName() function which actually outputs the various partition names. My driver write call syntax is as follows- getProvider->write(this, byteStart, WriteBuffer, completion); where this -> IOService* client, byteStart -> byte offset, WriteBuffer -> IOMemoryDescriptor*, completion -> IOStorageCompletion. But still my driver writes the buffer on all the media objects. How should we specify the write call of the driver to write the buffer on the specific media object? Or what parameters we have to pass to the write call, so that it is work for specific media object only? In our case the write call takes IOService* as input and the GetName() is IOMedia Method, so even if we check the name and fires the write call it will write on all the media objects. How should we give IOService* specific media object? Thanks Imran _______________________________________________ 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: host_basic_info
by Lassi A.Tuura 10 May '05

10 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, I know this isn't quite the answer you were looking for, but... http://developer.apple.com/qa/ov/ov01.html Lassi -- Next to knowing when to seize an opportunity, the most important thing in life is to know when to forego an advantage. --Benjamin Disraeli _______________________________________________ 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.… In Panther I used max_cpus to determine the max number of worker threads that should be spawned for a particular purpose. Does this equate to physical_cpu_max or logical_cpu_max in Darwin 8.0.0? This email sent to site_archiver(a)lists.apple.com

1 0

Re: host_basic_info
by Shaun Wexler 10 May '05

10 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On May 9, 2005, at 10:32 PM, William Kucharski wrote: On May 9, 2005, at 11:27 PM, Shaun Wexler wrote: development:~ shaun$ sysctl -a | grep cpu hw.ncpu = 2 hw.cpufrequency = 1416666661 hw.availcpu = 1 hw.ncpu: 2 hw.activecpu: 1 hw.physicalcpu: 1 hw.physicalcpu_max: 2 hw.logicalcpu: 1 hw.logicalcpu_max: 2 hw.cputype: 18 hw.cpusubtype: 11 hw.cpufrequency: 1416666661 hw.cpufrequency_min: 1416666661 hw.cpufrequency_max: 1416666661 _______________________________________________ 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.… In Panther I used max_cpus to determine the max number of worker threads that should be spawned for a particular purpose. Does this equate to physical_cpu_max or logical_cpu_max in Darwin 8.0.0? I need to use the largest of these values. I suppose that any *new* hardware will not run Panther anyway, but the question remains the same: which is largest, and which value will be used for max_cpus? I can't speak for Tiger, but based on experiences with other operating systems, logical_cpu_max will always be equal to or greater than physical_cpu_max. For example, given dual cores, given two physical CPUs you'd have four logical CPUs. Yes, that is also my assumption. Currently if I disable one CPU, it results in: What I want to know is if I should conditionalize on OS version, and read logical_cpu_max instead of max_cpus? Again, this is moot if dual-core machines will not support Panther, but we won't know that for a while... -- Shaun Wexler MacFOH http://www.macfoh.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: host_basic_info
by William Kucharski 10 May '05

10 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On May 9, 2005, at 11:27 PM, Shaun Wexler wrote: William Kucharski kucharsk(a)mac.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.… In Panther I used max_cpus to determine the max number of worker threads that should be spawned for a particular purpose. Does this equate to physical_cpu_max or logical_cpu_max in Darwin 8.0.0? I need to use the largest of these values. I suppose that any *new* hardware will not run Panther anyway, but the question remains the same: which is largest, and which value will be used for max_cpus? I can't speak for Tiger, but based on experiences with other operating systems, logical_cpu_max will always be equal to or greater than physical_cpu_max. For example, given dual cores, given two physical CPUs you'd have four logical CPUs. This email sent to site_archiver(a)lists.apple.com

1 0

host_basic_info
by Shaun Wexler 10 May '05

10 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.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 would like some clarification for the new fields in struct host_basic_info in Tiger, anticipating support for cpu's with multiple cores. struct host_basic_info { // PANTHER: integer_t max_cpus; /* max number of CPUs possible */ integer_t avail_cpus; /* number of CPUs now available */ natural_t memory_size; /* size of memory in bytes, capped at 2 GB */ cpu_type_t cpu_type; /* cpu type */ cpu_subtype_t cpu_subtype; /* cpu subtype */ // TIGER adds: cpu_threadtype_t cpu_threadtype; /* cpu threadtype */ integer_t physical_cpu; /* number of physical CPUs now available */ integer_t physical_cpu_max; /* max number of physical CPUs possible */ integer_t logical_cpu; /* number of logical cpu now available */ integer_t logical_cpu_max; /* max number of logical CPUs possible */ uint64_t max_mem; /* actual size of physical memory */ }; In Panther I used max_cpus to determine the max number of worker threads that should be spawned for a particular purpose. Does this equate to physical_cpu_max or logical_cpu_max in Darwin 8.0.0? I need to use the largest of these values. I suppose that any *new* hardware will not run Panther anyway, but the question remains the same: which is largest, and which value will be used for max_cpus? Also, I couldn't find anything except 0 for cpu_threadtype; are there other values of concern yet? -- Shaun Wexler MacFOH http://www.macfoh.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: how do I call soreceive?
by Jerry Cottingham 09 May '05

09 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On May 9, 2005, at 11:38 AM, Bijoy Thomas wrote: Hi, struct iovec aio; struct uio auio; aio.iov_base = &buf; //buf is a char[100] buf; aio.iov_len = 100; auio.uio_iov = &aio; auio.uio_iovcnt = 1; auio.uio_segflg = UIO_SYSSPACE; //we have a kernel space buffer auio.uio_rw = UIO_READ; auio.uio_offset = 0; auio.uio_resid = 100; auio.uio_procp = current_proc(); I am trying to understaned how to call soreceive from my kext. I am trying to receive data from an ftp server upon connect. My soconnect funtion returns 0 so I am assuming i have a good connect. I am trying to get the welcome message from the ftp server just for confirmation. The soreceive function looks fairly complex and my understanding is that it can do a variety of functions based on its supplied arguments. I just want to receive the welcome message. I do not want to pass it to user space. Upon browsing the kernel sources, I guess this can be accomplished by the following settings, and passing this so soreceive with both the mbufs as NULL. ( I'm guessing the mbufs are provided only if you want the detailed packet header and accompanying data returned possibly as a list of mbufs). I tried it this way but i'm getting a non zero return from soreceive. Any suggestions/links? I can't offer any help with soreceive, but you really should avoid accessing the uio structure directly. Use the new uio KPI if at all possible (see uio.h in Tiger). _______________________________________________ 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: how do I call soreceive?
by Josh Graessley 09 May '05

09 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On May 9, 2005, at 1:13 PM, Bijoy Thomas wrote: <snip> -josh _______________________________________________ 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 If we are using a tcp socket, do we need to provide the second arg for soreceive or is that only for udp sockets. I think it's optional. The source code is available, so if you're really curious you can dig through the function yourself. It's a little complex, but not nearly as daunting as sosend ;) Am i correct in my assumtion that as long as i care only about the data, i just need to provide the struct uio arg for sorecieve with its iovec filed initialzed to my kernel space buffer and its length and setting uio_segflg = UIO_SYSSPACE and uio_rw = UIO_READ. In this case the mbufs can be NULL in the call to soreceive. If you fill out the uio with information on where to store the data, you can leave the mbuf NULL. If you want to receive the data in an mbuf chain instead of having the data copied in to your buffer, you could do this by just specifying the uio_resid and fill out the mbuf parameter. Also, why do we use the uio_resid field in the struct iuo? Is it used in multiple calls. Can we just set it to the buffer length on the first call? You can use this value to determine how much data was received. Set it to the maximum size of your buffers when you call soreceive. Keep a copy of what that value was. Upon return, you can retrieve the value again then subtract that from the original to determine the amount of data received. This value is also useful for limiting the maximum amount of data to receive in an mbuf chain. I was using the flag MSG_WAITALL for the soreceive. Will that cause the soreceive to stall until all data arrives? Is there a default flag for just returning whatever data has arrived. soreceive behaves the same as recvmsg in userspace with respect to waiting. smime.p7s

1 0

Re: how do I call soreceive?
by Bijoy Thomas 09 May '05

09 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, Thanks in advance Bijoy Thomas. On May 9, 2005, at 2:49 PM, Justin Walker wrote: On May 9, 2005, at 11:38, Bijoy Thomas wrote: Regards, Justin -- Justin C. Walker, Curmudgeon-at-Large /~\ The ASCII \ / Ribbon Campaign X Help Cure HTML Email / \ This email sent to bijoyjth(a)bluebottle.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.… Thanks for all your suggestions. I made the wrong asusmption that return values were negative .. so i was doing if(soconnect(args) < 0){ printf("bad socket"); } when actually soconect was returning EINVAL (bad arg) because i forgot to fill the sin_len field of my sockaddr_in argument. Hence, the ensuing soreceive was returning ENOTCONN (not connected). Now thats fixed and my soconnect returns 0. The problem now is that my userland program seems to be sleeping because after hitting ctrl-C, the return value from soreceive is EINTR. Hence, i guess my program is blocked by soreceive. I have a couple of questions If we are using a tcp socket, do we need to provide the second arg for soreceive or is that only for udp sockets. Am i correct in my assumtion that as long as i care only about the data, i just need to provide the struct uio arg for sorecieve with its iovec filed initialzed to my kernel space buffer and its length and setting uio_segflg = UIO_SYSSPACE and uio_rw = UIO_READ. In this case the mbufs can be NULL in the call to soreceive. Also, why do we use the uio_resid field in the struct iuo? Is it used in multiple calls. Can we just set it to the buffer length on the first call? I was using the flag MSG_WAITALL for the soreceive. Will that cause the soreceive to stall until all data arrives? Is there a default flag for just returning whatever data has arrived. I know I am jumping ahead of myself without understanding the workings of soreceive. I have alredy placed an order for the design of the bsd os book. But until then, I would be grateful if anyone could answer these. [snip] and passing this so soreceive with both the mbufs as NULL. ( I'm guessing the mbufs are provided only if you want the detailed packet header and accompanying data returned possibly as a list of mbufs). I tried it this way but i'm getting a non zero return from soreceive. Any suggestions/links? My first suggestion is to actually look at, and understand, the non-zero return value you get from soreceive(). If you don't understand the value, it might help to let us know what it is. _______________________________________________ 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/ bijoyjth%40bluebottle.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Problems building xnu-792
by Markus Hitter 09 May '05

09 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com To answer a second time ... Am 09.05.2005 um 05:49 schrieb William Kucharski: This specific line translates to this line in osfmk/ppc/hw_vm.s: rlwinm r12,r11,GV_HPAGE_SHIFT,GV_HPAGE_MASK which SHOULD assemble to: rlwinm r12,r11,31,23,28 Cheers, Markus - - - - - - - - - - - - - - - - - - - Dipl. Ing. Markus Hitter http://www.jump-ing.de/ _______________________________________________ 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.… according to the 10.4 mach_kernel. [...] #define GV_HPAGE_SHIFT (GV_PGIDX_SZ_LG2 - GV_GRPS_PPG_LG2) [...] calculates out as (3 - (12 - 5 + 3)))), or -1. Oooh. After finally diving into Motorola's multi-thousand-pages PPC manuals, I found rlwinm to be a bit shift operator and the third parameter is the number of bits to be shifted. This is a 5-bit parameter, so 31 == -1. It's a compiler missing feature/bug. After redefining GV_HPAGE_SHIFT in osfmk/ppc/mappings.h to 31 (works for now) and a "make clean; make" the compile completed. Two missing symbols in the final link stage left for enjoyment tomorrow. This email sent to site_archiver(a)lists.apple.com

1 0