)]}' { "log": [ { "commit": "958b166c00b39ff0b28ad2bbb32624b9f305a4e1", "tree": "9a7d086f44e4fc034f1f91eed945ef4eabc5fa0c", "parents": [ "2db8d99ffdbed7d2beb1bbdefdcd086dda9dee98" ], "author": { "name": "Keith Owens", "email": "kaos@sgi.com", "time": "Mon Apr 03 15:26:12 2006 +1000" }, "committer": { "name": "Tony Luck", "email": "tony.luck@intel.com", "time": "Fri Apr 07 22:51:51 2006 -0700" }, "message": "[IA64] Pass more data to the MCA/INIT notify_die hooks\n\nThe MCA/INIT handlers maintain important state in the SAL to OS (sos)\narea and in the monarch_cpu flag. Kernel debuggers (such as KDB) need\nthis data, and may need to adjust the monarch_cpu field so make the\ndata available to the notify_die hooks. Define two more events for\ncalling the functions on the notify_die chain.\n\nSigned-off-by: Keith Owens \u003ckaos@sgi.com\u003e\nSigned-off-by: Tony Luck \u003ctony.luck@intel.com\u003e\n" }, { "commit": "e041c683412d5bf44dc2b109053e3b837b71742d", "tree": "9d271066ef379da0c0fb3b8cb4137abd5d2ebba0", "parents": [ "76b81e2b0e2241accebcc68e126bc5ab958661b9" ], "author": { "name": "Alan Stern", "email": "stern@rowland.harvard.edu", "time": "Mon Mar 27 01:16:30 2006 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Mon Mar 27 08:44:50 2006 -0800" }, "message": "[PATCH] Notifier chain update: API changes\n\nThe kernel\u0027s implementation of notifier chains is unsafe. There is no\nprotection against entries being added to or removed from a chain while the\nchain is in use. The issues were discussed in this thread:\n\n http://marc.theaimsgroup.com/?l\u003dlinux-kernel\u0026m\u003d113018709002036\u0026w\u003d2\n\nWe noticed that notifier chains in the kernel fall into two basic usage\nclasses:\n\n\t\"Blocking\" chains are always called from a process context\n\tand the callout routines are allowed to sleep;\n\n\t\"Atomic\" chains can be called from an atomic context and\n\tthe callout routines are not allowed to sleep.\n\nWe decided to codify this distinction and make it part of the API. Therefore\nthis set of patches introduces three new, parallel APIs: one for blocking\nnotifiers, one for atomic notifiers, and one for \"raw\" notifiers (which is\nreally just the old API under a new name). New kinds of data structures are\nused for the heads of the chains, and new routines are defined for\nregistration, unregistration, and calling a chain. The three APIs are\nexplained in include/linux/notifier.h and their implementation is in\nkernel/sys.c.\n\nWith atomic and blocking chains, the implementation guarantees that the chain\nlinks will not be corrupted and that chain callers will not get messed up by\nentries being added or removed. For raw chains the implementation provides no\nguarantees at all; users of this API must provide their own protections. (The\nidea was that situations may come up where the assumptions of the atomic and\nblocking APIs are not appropriate, so it should be possible for users to\nhandle these things in their own way.)\n\nThere are some limitations, which should not be too hard to live with. For\natomic/blocking chains, registration and unregistration must always be done in\na process context since the chain is protected by a mutex/rwsem. Also, a\ncallout routine for a non-raw chain must not try to register or unregister\nentries on its own chain. (This did happen in a couple of places and the code\nhad to be changed to avoid it.)\n\nSince atomic chains may be called from within an NMI handler, they cannot use\nspinlocks for synchronization. Instead we use RCU. The overhead falls almost\nentirely in the unregister routine, which is okay since unregistration is much\nless frequent that calling a chain.\n\nHere is the list of chains that we adjusted and their classifications. None\nof them use the raw API, so for the moment it is only a placeholder.\n\n ATOMIC CHAINS\n -------------\narch/i386/kernel/traps.c:\t\ti386die_chain\narch/ia64/kernel/traps.c:\t\tia64die_chain\narch/powerpc/kernel/traps.c:\t\tpowerpc_die_chain\narch/sparc64/kernel/traps.c:\t\tsparc64die_chain\narch/x86_64/kernel/traps.c:\t\tdie_chain\ndrivers/char/ipmi/ipmi_si_intf.c:\txaction_notifier_list\nkernel/panic.c:\t\t\t\tpanic_notifier_list\nkernel/profile.c:\t\t\ttask_free_notifier\nnet/bluetooth/hci_core.c:\t\thci_notifier\nnet/ipv4/netfilter/ip_conntrack_core.c:\tip_conntrack_chain\nnet/ipv4/netfilter/ip_conntrack_core.c:\tip_conntrack_expect_chain\nnet/ipv6/addrconf.c:\t\t\tinet6addr_chain\nnet/netfilter/nf_conntrack_core.c:\tnf_conntrack_chain\nnet/netfilter/nf_conntrack_core.c:\tnf_conntrack_expect_chain\nnet/netlink/af_netlink.c:\t\tnetlink_chain\n\n BLOCKING CHAINS\n ---------------\narch/powerpc/platforms/pseries/reconfig.c:\tpSeries_reconfig_chain\narch/s390/kernel/process.c:\t\tidle_chain\narch/x86_64/kernel/process.c\t\tidle_notifier\ndrivers/base/memory.c:\t\t\tmemory_chain\ndrivers/cpufreq/cpufreq.c\t\tcpufreq_policy_notifier_list\ndrivers/cpufreq/cpufreq.c\t\tcpufreq_transition_notifier_list\ndrivers/macintosh/adb.c:\t\tadb_client_list\ndrivers/macintosh/via-pmu.c\t\tsleep_notifier_list\ndrivers/macintosh/via-pmu68k.c\t\tsleep_notifier_list\ndrivers/macintosh/windfarm_core.c\twf_client_list\ndrivers/usb/core/notify.c\t\tusb_notifier_list\ndrivers/video/fbmem.c\t\t\tfb_notifier_list\nkernel/cpu.c\t\t\t\tcpu_chain\nkernel/module.c\t\t\t\tmodule_notify_list\nkernel/profile.c\t\t\tmunmap_notifier\nkernel/profile.c\t\t\ttask_exit_notifier\nkernel/sys.c\t\t\t\treboot_notifier_list\nnet/core/dev.c\t\t\t\tnetdev_chain\nnet/decnet/dn_dev.c:\t\t\tdnaddr_chain\nnet/ipv4/devinet.c:\t\t\tinetaddr_chain\n\nIt\u0027s possible that some of these classifications are wrong. If they are,\nplease let us know or submit a patch to fix them. Note that any chain that\ngets called very frequently should be atomic, because the rwsem read-locking\nused for blocking chains is very likely to incur cache misses on SMP systems.\n(However, if the chain\u0027s callout routines may sleep then the chain cannot be\natomic.)\n\nThe patch set was written by Alan Stern and Chandra Seetharaman, incorporating\nmaterial written by Keith Owens and suggestions from Paul McKenney and Andrew\nMorton.\n\n[jes@sgi.com: restructure the notifier chain initialization macros]\nSigned-off-by: Alan Stern \u003cstern@rowland.harvard.edu\u003e\nSigned-off-by: Chandra Seetharaman \u003csekharan@us.ibm.com\u003e\nSigned-off-by: Jes Sorensen \u003cjes@sgi.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "9138d581b0ef855c0314c41c14852a7231b9941c", "tree": "f8aed7413d873877bc8373a37055308d388faa06", "parents": [ "5b2f7ffcb734d3046144dfbd5ac6d76254a9e522" ], "author": { "name": "Keith Owens", "email": "kaos@sgi.com", "time": "Mon Nov 07 11:27:13 2005 -0800" }, "committer": { "name": "Tony Luck", "email": "tony.luck@intel.com", "time": "Mon Nov 07 11:27:13 2005 -0800" }, "message": "[IA64] Extend notify_die() hooks for IA64\n\nnotify_die() added for MCA_{MONARCH,SLAVE,RENDEZVOUS}_{ENTER,PROCESS,LEAVE} and\nINIT_{MONARCH,SLAVE}_{ENTER,PROCESS,LEAVE}. We need multiple\nnotification points for these events because they can take many seconds\nto run which has nasty effects on the behaviour of the rest of the\nsystem.\n\nDIE_SS replaced by a generic DIE_FAULT which checks the vector number,\nto allow interception of faults other than SS.\n\nDIE_MACHINE_{HALT,RESTART} added to allow last minute close down\nprocessing, especially when the halt/restart routines are called from\nerror handlers.\n\nDIE_OOPS added.\n\nThe check for kprobe\u0027s break numbers has been moved from traps.c to\nkprobes.c, allowing DIE_BREAK to be used for any additional break\nnumbers, i.e. it is no longer kprobes specific.\n\nHooks for kernel debuggers and kernel dumpers added, ENTER and LEAVE.\nBoth of these disable the system for long periods which impact on\nwatchdogs and heartbeat systems in general. More patches to come that\nuse these events to reset watchdogs and heartbeats.\n\nunregister_die_notifier() added and both routines exported. Requested\nby Dean Nelson.\n\nLock removed from {un,}register_die_notifier. notifier_chain_register()\nalready takes a lock. Also the generic notifier chain locking is being\nreworked to distinguish between callbacks that can block and those that\ncannot, the lock in {un,}register_die_notifier would interfere with\nthat change. http://marc.theaimsgroup.com/?l\u003dlinux-kernel\u0026m\u003d113018709002036\u0026w\u003d2\n\nLeading white space removed from arch/ia64/kernel/kprobes.c.\n\nTypo in mca.c in original version of this patch found \u0026 fixed by Dean\nNelson.\n\nSigned-off-by: Keith Owens \u003ckaos@sgi.com\u003e\nAcked-by: Dean Nelson \u003cdcn@sgi.com\u003e\nAcked-by: Anil Keshavamurthy \u003canil.s.keshavamurthy@intel.com\u003e\nSigned-off-by: Tony Luck \u003ctony.luck@intel.com\u003e\n" }, { "commit": "7213b2521889eb087eed8abaa48d1a692575da3e", "tree": "8a9a0b5cfaa9824de97d9dae45b20d2a7309db5b", "parents": [ "0aa55e4d7db822059fe8132fe9f2b7773c48216c" ], "author": { "name": "Anil S Keshavamurthy", "email": "anil.s.keshavamurthy@intel.com", "time": "Thu Jun 23 00:09:27 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Thu Jun 23 09:45:22 2005 -0700" }, "message": "[PATCH] Kprobes/IA64: kdebug die notification mechanism\n\nAs many of you know that kprobes exist in the main line kernel for various\narchitecture including i386, x86_64, ppc64 and sparc64. Attached patches\nfollowing this mail are a port of Kprobes and Jprobes for IA64.\n\nI have tesed this patches for kprobes and Jprobes and this seems to work fine.\n I have tested this patch by inserting kprobes on various slots and various\ntemplates including various types of branch instructions.\n\nI have also tested this patch using the tool\nhttp://marc.theaimsgroup.com/?l\u003dlinux-kernel\u0026m\u003d111657358022586\u0026w\u003d2 and the\nkprobes for IA64 works great.\n\nHere is list of TODO things and pathes for the same will appear soon.\n\n1) Support kprobes on \"mov r1\u003dip\" type of instruction\n2) Support Kprobes and Jprobes to exist on the same address\n3) Support Return probes\n3) Architecture independent cleanup of kprobes\n\nThis patch adds the kdebug die notification mechanism needed by Kprobes.\n\nFor break instruction on Branch type slot, imm21 is ignored and value\nzero is placed in IIM register, hence we need to handle kprobes\nfor switch case zero.\n\nSigned-off-by: Anil S Keshavamurthy \u003canil.s.keshavamurthy@intel.com\u003e\nSigned-off-by: Rusty Lynch \u003cRusty.lynch@intel.com\u003e\n\nFrom: Rusty Lynch \u003crusty.lynch@intel.com\u003e\n\nAt the point in traps.c where we recieve a break with a zero value, we can\nnot say if the break was a result of a kprobe or some other debug facility.\n\nThis simple patch changes the informational string to a more correct \"break\n0\" value, and applies to the 2.6.12-rc2-mm2 tree with all the kprobes\npatches that were just recently included for the next mm cut.\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" } ] }