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

Re: Where is the "REAL" kernel of my G5?
by Markus Hitter 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Am 11.05.2005 um 19:29 schrieb Ramble Leon: Type of hardware shouldn't matter as long as it's supported. ... the 1.6UP is working fine, the 1.8DP is also bootable, but the booting kernel is not my kernel. HTH, 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.… I have two G5, one is 1.6G UP, another is 1.8G DP, ... ... I've made my own kernel and put xnu/BUILD/dst/* ... Copying the file named "mach_kernel" to the boot volumes' root directory is sufficient. Then you obviously have other bootable kernels or NetBoot available or might not boot the volume you think. Did you perhaps just rename the old kernel instead of copying & deleting it? Perhaps, modern versions of BootX remember something like the file's inode instead of the file path ... Where is the REAL kernel in the 1.8DP G5? How can I replace it? It's /mach_kernel, loaded into memory by BootX. I'd go to the Startup Volume system panel and check what's available. Then, click on the volume you want to boot to make the panel re-write OpenFirmware settings. If this doesn't help, you'd have to fiddle with the "nvram" utility directly, but I doubt that. This email sent to site_archiver(a)lists.apple.com

1 0

Where is the "REAL" kernel of my G5?
by Ramble Leon 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; b=OeByZr+XK27QUt2uUtapGc56P2MXuc5OUiPFoHXRmDa1Ba7RugThgsNSVwDhK+G6TVKhIm9JzN4UPcryi3YMsQ6JgvhlaojuSGNIBJDPmadg+s0YpDkbFhhZ8HKZHzw3Qou2o7sFkY9CaMIhu+JbZ2CJ01eirtQ61MWlBU70WnY= ; I have two G5, one is 1.6G UP, another is 1.8G DP, I've made my own kernel and put xnu/BUILD/dst/* in a pkg, by install the package, the mach_kernel will be copied to /mach_kernel, kexts and frameworks will be copied to /System/Library/Extensions & Frameworks. I installed the package in both my G5, the 1.6UP is working fine, the 1.8DP is also bootable, but the booting kernel is not my kernel. Although I check the /mach_kernel in both machines by md5, they are exactly same kernel, but uname of these two machines are different, uname output of the 1.8DP is information of the old kernel. Even I removed the old one and reboot, still the same result. Where is the REAL kernel in the 1.8DP G5? How can I replace it? Anybody can help me a little bit? Thanks. __________________________________ Yahoo! Mail Mobile Take Yahoo! Mail with you! Check email on your mobile phone. http://mobile.yahoo.com/learn/mail _______________________________________________ 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: argument defs for vnode scope callbacks
by Quinn 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com One more question, though : Is it possible to have more specific informations about the auth request, like the real VNOP that was issued and which generated the request ? Not as far as I know. You should have seen a write authorization request for the parent directory. One last thing : You say that "There are many other things you can use with VATTR_WANTED." #define VNODE_ATTR_va_rdev (1LL<< 0) /* 00000001 */ #define VNODE_ATTR_va_nlink (1LL<< 1) /* 00000002 */ #define VNODE_ATTR_va_total_size (1LL<< 2) /* 00000004 */ #define VNODE_ATTR_va_total_alloc (1LL<< 3) /* 00000008 */ #define VNODE_ATTR_va_data_size (1LL<< 4) /* 00000010 */ #define VNODE_ATTR_va_data_alloc (1LL<< 5) /* 00000020 */ #define VNODE_ATTR_va_iosize (1LL<< 6) /* 00000040 */ #define VNODE_ATTR_va_uid (1LL<< 7) /* 00000080 */ #define VNODE_ATTR_va_gid (1LL<< 8) /* 00000100 */ #define VNODE_ATTR_va_mode (1LL<< 9) /* 00000200 */ #define VNODE_ATTR_va_flags (1LL<<10) /* 00000400 */ #define VNODE_ATTR_va_acl (1LL<<11) /* 00000800 */ #define VNODE_ATTR_va_create_time (1LL<<12) /* 00001000 */ #define VNODE_ATTR_va_access_time (1LL<<13) /* 00002000 */ #define VNODE_ATTR_va_modify_time (1LL<<14) /* 00004000 */ #define VNODE_ATTR_va_change_time (1LL<<15) /* 00008000 */ #define VNODE_ATTR_va_backup_time (1LL<<16) /* 00010000 */ #define VNODE_ATTR_va_fileid (1LL<<17) /* 00020000 */ #define VNODE_ATTR_va_linkid (1LL<<18) /* 00040000 */ #define VNODE_ATTR_va_parentid (1LL<<19) /* 00080000 */ #define VNODE_ATTR_va_fsid (1LL<<20) /* 00100000 */ #define VNODE_ATTR_va_filerev (1LL<<21) /* 00200000 */ #define VNODE_ATTR_va_gen (1LL<<22) /* 00400000 */ #define VNODE_ATTR_va_encoding (1LL<<23) /* 00800000 */ #define VNODE_ATTR_va_type (1LL<<24) /* 01000000 */ #define VNODE_ATTR_va_name (1LL<<25) /* 02000000 */ #define VNODE_ATTR_va_uuuid (1LL<<26) /* 04000000 */ #define VNODE_ATTR_va_guuid (1LL<<27) /* 08000000 */ #define VNODE_ATTR_va_nchildren (1LL<<28) /* 10000000 */ S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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 14:55 +0200 11/5/05, Nicolas Berloquin wrote: vnode_if.h lists all kinds of VNOPs that seem to be more finegrained than the bits passed to the callback. I logged the bitfield for each call, and I wasn't able to see when a file gets created for example... However, I should stress that (apart from for the KATH_SCOPE_FILEOP), Kauth is not a notification mechanism, it's an authorization mechanism. Hence you get information about the action being authorized, not the higher-level operation that caused it. Is this completely undocumented, or did I miss a piece of doc ? It's not documented, per se, but there's a list in the <sys/vnode.h> header in the Kernel framework. If you look at the VATTR_WANTED macro, you'll see that it ends up referencing these constants by concatenating VNODE_ATTR_ and its argument (eg va_nchildren). This email sent to site_archiver(a)lists.apple.com

1 0

Re: argument defs for vnode scope callbacks
by Nicolas Berloquin 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thanks ! That looks exactly like what I needed. Is this completely undocumented, or did I miss a piece of doc ? thanks ;-D On May 11, 2005, at 1:29 PM, John Dalgliesh wrote: Hi, I didn't see a reply to this one yet, so here's my attempt... On Tue, 10 May 2005, Nicolas Berloquin wrote: I then tried to get some attributes from the vp vnode, but I got instant-KP out of it : struct vnode_attr vap; int err = vnode_getattr(vp, &vap, ctx); any suggestions are welcome. struct vnode_attr vap; VATTR_INIT( &vap ); VATTR_WANTED( &vap, va_data_size ); int err = vnode_getattr( vp, &vap, ctx ); There are many other things you can use with VATTR_WANTED. Hope that helps! {P^/ _______________________________________________ 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.… One more question, though : Is it possible to have more specific informations about the auth request, like the real VNOP that was issued and which generated the request ? vnode_if.h lists all kinds of VNOPs that seem to be more finegrained than the bits passed to the callback. I logged the bitfield for each call, and I wasn't able to see when a file gets created for example... One last thing : You say that "There are many other things you can use with VATTR_WANTED." The vnode_attr thingy must first be initialised, and then told what you want to get out of it. e.g. if you were interested in the size of the data fork, you'd do this: This email sent to site_archiver(a)lists.apple.com

1 0

RE: Read through Filter Scheme Driver
by Imran Kazi 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcVVmyk7tDtJYOR/TEyQM0gVRWbn6wAigEHQ Dear Mike, Please go through the below code for our Read issue /***************User Space Application **************************/ /* Purpose: This calls the 'read' method of user client using IOConnectMethodStructureIStructureO */ static kern_return_t MyDoRead(io_connect_t connect) kern_return_t kernResult; MySampleStruct sampleStructIn; // Input structure MySampleStruct sampleStructOut; //Output structure kernResult = vm_allocate(mach_task_self(), (vm_address_t*) &sampleStructIn.buff, vm_page_size, TRUE); if(sampleStructIn.buff) { char* buff = (char*) sampleStructOut.buff; strcpy(buff, ""); } IOByteCount structSizeIn = sizeof(MySampleStruct); IOByteCount structSizeOut = sizeof(MySampleStruct); if (connect) { kernResult = IOConnectMethodStructureIStructureO(connect, // connection handle returned from IOServiceOpen(). kCoreDoRead, // an index to the function in the kernel. structSizeIn, // the number of scalar parameters. &structSizeOut, &sampleStructIn, &sampleStructOut ); } return kernResult; } /***************User client **************************/ getTargetAndMethodForIndex(IOService ** target, UInt32 index) { { // kCoreDoRead NULL, (IOMethod) &com_CoreSnap_UserClient::CoreSnapUserClientDoRead, kIOUCStructIStructO, sizeof(MySampleStruct), sizeof(MySampleStruct) }, } IOReturn com_CoreSnap_UserClient::CoreSnapUserClientDoRead(MySampleStruct *InStruct, IOByteCount InStructSize, MySampleStruct *OutStruct, IOByteCount *OutStructSize) { IOMemoryDescriptor *ReadBuffer = NULL; // Constructing memory descriptor for out put structure ReadBuffer = IOMemoryDescriptor::withAddress((vm_address_t) OutStruct->buff, (IOByteCount) OutStruct->size, kIODirectionIn, fTask); if(ReadBuffer) //Created IOMemoryDescriptor for buff Successfully.... { retRead = ReadBuffer->prepare(); IOLog("\n address of prepare buffer = %p\n", retRead); if(retRead == kIOReturnSuccess) { IOLog("\n prepare for buff Succeded....\n"); // mapping the IOMemoryDescriptr buffer retRead = ReadBuffer->map(); if(retRead == kIOReturnSuccess) { IOLog("map successful\n"); // get the provider for our userclient and call driver read routine UCProvider = OSDynamicCast(com_CoreSnap_driver, fProvider); IOLog("\n UCProvider = %p\n", UCProvider); IOStorageCompletion completion; // calling driver read call by passing IOMemoryDescriptor buffer retRead = UCProvider->CoreSnapDriverDoRead(this, 512, ReadBuffer, completion); if(retRead == kIOReturnSuccess) { IOLog("\n CoreSnapDriverDoRead of Driver Succeded...\n"); //return retRead; } }// end of prepare } ReadBuffer->complete(); ReadBuffer->release(); if (kIOReturnSuccess != retRead) { IOLog("%s[%p]::%s - returning err 0x%08x", getName(), this, __FUNCTION__, retRead); } fProvider->release(); return retRead; } /********************************Driver Read routine ***************************/ // kCoreDoRead IOReturn com_CoreSnap_driver::CoreSnapDriverDoRead(IOService* client, UInt64 byteStart, IOMemoryDescriptor * OutBuffer, IOStorageCompletion completion) { IOReturn retDrRead = kIOReturnSuccess; IOByteCount readBytes; IOByteCount requestByteCount = 0; IOReturn status = 0; IOByteCount bufLength = OutBuffer->getLength(); char *myBuff = new char[bufLength]; readBytes = OutBuffer->readBytes(0, myBuff, bufLength); IOLog("com_CoreSnap_driver::read BEFORE **** readBytes = %lu byteStart = %lu \ bufLength = %lu buffer = %s ***\n", (long)readBytes, (long)byteStart, (long)bufLength, myBuff); // Make "filter" changes to the buffer here. IOLog("\n Calling read routine of driver.....\n"); getProvider()->read(client, 0, OutBuffer, completion); IOLog("\n Called read routine of driver.....\n"); IOLog("\n Calling readbytes.......in driver\n"); readBytes = OutBuffer->readBytes(0, myBuff, bufLength); IOLog("com_CoreSnap_driver::***AFTER::readBytes = %lu byteStart = %lu bufLength = %lu buffer = %s \n", (long)readBytes, (long)byteStart, (long)bufLength, myBuff); IOLog("\n Calling complete routine of driver......\n"); getProvider()->complete(completion, status, requestByteCount); IOLog("\n Called complete routine of driver......\n"); delete[] myBuff; return retDrRead; } We are able to prepare() and map() IOMemoryDescriptor successfully and pass this buffer to driver's read routine(we also have interchange the prepare and map). My problem is while debugging it prints all the IOLogs before driver's standard read() method, but IOLogs after read() does not get printed in system.logs. Do we need to call withAddress() and prepare() and map() the IOMemoryDescriptor before passing it to standard read() method. Are we following correct steps for reading the buffer from disk and passing it to user space application? 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: Read through Filter Scheme Driver
by Imran Kazi 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcVVmyk7tDtJYOR/TEyQM0gVRWbn6wAiYKSA _______________________________________________ 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: argument defs for vnode scope callbacks
by John Dalgliesh 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi, I didn't see a reply to this one yet, so here's my attempt... On Tue, 10 May 2005, Nicolas Berloquin wrote: > > I then tried to get some attributes from the vp vnode, but I got > instant-KP out of it : > > struct vnode_attr vap; > int err = vnode_getattr(vp, &vap, ctx); > > any suggestions are welcome. The vnode_attr thingy must first be initialised, and then told what you want to get out of it. e.g. if you were interested in the size of the data fork, you'd do this: struct vnode_attr vap; VATTR_INIT( &vap ); VATTR_WANTED( &vap, va_data_size ); int err = vnode_getattr( vp, &vap, ctx ); There are many other things you can use with VATTR_WANTED. Hope that helps! {P^/ _______________________________________________ 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

VFS Tidbits
by Quinn 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Greetings All * * * vfs_statfs is a simple accessor function. Here's the Tiger implementation. struct vfsstatfs * vfs_statfs(mount_t mp) { return(&mp->mnt_vfsstat); } * * * void vfs_event_signal( __unused fsid_t * fsid, u_int32_t event, __unused intptr_t data ) { KNOTE(&fs_klist, event); } * * * Q: What does 64-bit ready (VFS_TBL64BITREADY) mean? o mount o ioctl o sysctl * * * Share and Enjoy -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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.… As part of my 'day job' <http://developer.apple.com/technicalsupport/index.html> I answer questions from developers who are porting their file systems to Tiger. As I come up with interesting tidbits about Tiger's VFS KPI, I'll post it to the list. Obviously the lifetime of the pointer to the returned structure is based on the lifetime of the mountpoint itself. vfs_event_signal is used to generate kqueue events for the volume. The implementation is /super/ trivial: However, that's not exactly helpful. What you really want to know is under what circumstances you should call it and what the parameters are. I'll start with the second question first. o fsid is the BSD FSID of your volume. If you don't have this handy, you can get it using vfs_statfs. o data is event specific data. AFAIK none of the currently defined events have any associated data (the __unused in the code above is a dead giveaway), so you should just pass NULL. o event is a bitmap of the new events. You can see the definitions in <sys/mount.h>. This list of events answers the question as to when you should call vfs_event_signal. You should call it when one of these events occurs. This may or may not happen, depending on the specifics of your file system. For example, HFS only ever sends the VQ_LOWDISK event. However, network file systems might generate a whole raft of other events. A: This flag tells VFS whether your file system is prepared to handle certain requests from a 64-bit processes. These include: If this flag isn't set and the call comes from a 64-bit process, VFS will fail the call. This prevents you from accidentally interpreting a 64-bit structure (such as the mount arguments) as a 32-bit structure. If you set the 64-bit ready flag, you're expected to test whether the caller is a 64-bit process and interpret the structure appropriately. This email sent to site_archiver(a)lists.apple.com

1 0

Re: syscall implementation
by Quinn 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Cool. Exactly. Share and Enjoy -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ 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 19:02 -0400 10/5/05, Bijoy Thomas wrote: I was just curious about how they worked on the ppc. By system call patching, do you mean that the syscall table is no longer exported and replacing system calls with your own functions or trying to add a new syscall is no longer possible? I remember the Linux kernel developers got tired of people hijacking syscalls and decided not to export it any more with version 2.6. But ofcourse, people have found ways around it. Of course. Mac OS X is an open source kernel, after all, so we can't stop folks from hacking it. However, as we draw a sharp line between what's supported (the KPIs) and what's not (everything else), we feel a lot less worried about changing the unsupported stuff and breaking the developers that rely on it. The flip side of this coin is that we're serious about extending the KPIs to cover serious developer requirements. So, if you're building a product that requires you to patch the system call dispatch table, we'd like to know about it. This happened a number of times during the Tiger pre-release cycle, and I expect it to happen again as we (Apple and our third party developers) get more experience developing software for Tiger. This email sent to site_archiver(a)lists.apple.com

1 0

Re: sbdrop - count not zero?
by Stephane Sudre 11 May '05

11 May '05

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On 11 mai 05, at 02:17, Josh Graessley wrote: Yes, there was/is one which is registered with these parameters: _______________________________________________ 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.… Stephane Sudre wrote a question about sbdrop. There is a known bug where this may occur if there is a socket filter installed and it is filtering inbound data. Stephane, were you running any socket filters (kernel extensions)? tStatus = sflt_register(&SFM_SocketFilter_TCP, PF_INET,SOCK_STREAM, IPPROTO_TCP); and which had indeed an inbound data callback (the return value is always 0). This email sent to site_archiver(a)lists.apple.com

1 0