)]}' { "log": [ { "commit": "51c6f666fceb3184eeff045dad4432b602cd648e", "tree": "33e29916e0fea872ba6f29eba698219a740b078f", "parents": [ "885036d32f5d3c427c3e2b385b5a5503805e3e52" ], "author": { "name": "Robin Holt", "email": "holt@sgi.com", "time": "Sun Nov 13 16:06:42 2005 -0800" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sun Nov 13 18:14:12 2005 -0800" }, "message": "[PATCH] mm: ZAP_BLOCK causes redundant work\n\nThe address based work estimate for unmapping (for lockbreak) is and always\nwas horribly inefficient for sparse mappings. The problem is most simply\nexplained with an example:\n\nIf we find a pgd is clear, we still have to call into unmap_page_range\nPGDIR_SIZE / ZAP_BLOCK_SIZE times, each time checking the clear pgd, in\norder to progress the working address to the next pgd.\n\nThe fundamental way to solve the problem is to keep track of the end\naddress we\u0027ve processed and pass it back to the higher layers.\n\nFrom: Nick Piggin \u003cnpiggin@suse.de\u003e\n\n Modification to completely get away from address based work estimate\n and instead use an abstract count, with a very small cost for empty\n entries as opposed to present pages.\n\n On 2.6.14-git2, ppc64, and CONFIG_PREEMPT\u003dy, mapping and unmapping 1TB\n of virtual address space takes 1.69s; with the following patch applied,\n this operation can be done 1000 times in less than 0.01s\n\nFrom: Andrew Morton \u003cakpm@osdl.org\u003e\n\nWith CONFIG_HUTETLB_PAGE\u003dn:\n\nmm/memory.c: In function `unmap_vmas\u0027:\nmm/memory.c:779: warning: division by zero\n\nDue to\n\n\t\t\tzap_work -\u003d (end - start) /\n\t\t\t\t\t(HPAGE_SIZE / PAGE_SIZE);\n\nSo make the dummy HPAGE_SIZE non-zero\n\nSigned-off-by: Robin Holt \u003cholt@sgi.com\u003e\nSigned-off-by: Nick Piggin \u003cnpiggin@suse.de\u003e\nCc: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "1a44e149084d772a1bcf4cdbdde8a013a8a1cfde", "tree": "b3f682ce8df89edb9740fdd5c178df5accc49736", "parents": [ "2e9b367c2273ed21c9852a04d90944d472c4f3e6" ], "author": { "name": "Andrea Arcangeli", "email": "andrea@suse.de", "time": "Sat Oct 29 18:16:48 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:43 2005 -0700" }, "message": "[PATCH] .text page fault SMP scalability optimization\n\nWe had a problem on ppc64 where with more than 4 threads a large system\nwouldn\u0027t scale well while faulting in the .text (most of the time was spent\nin the kernel despite it was an userland compute intensive app). The\nreason is the useless overwrite of the same pte from all cpu.\n\nI fixed it this way (verified on an older kernel but the forward port is\nalmost identical). This will benefit all archs not just ppc64.\n\nSigned-off-by: Andrea Arcangeli \u003candrea@suse.de\u003e\nCc: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "f412ac08c9861b4791af0145934c22f1458686da", "tree": "5e515efa116f3968c2caa75bc691a197199313a8", "parents": [ "4c21e2f2441dc5fbb957b030333f5a3f2d02dea7" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:41 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:42 2005 -0700" }, "message": "[PATCH] mm: fix rss and mmlist locking\n\nA couple of oddities were guarded by page_table_lock, no longer properly\nguarded when that is split.\n\nThe mm_counters of file_rss and anon_rss: make those an atomic_t, or an\natomic64_t if the architecture supports it, in such a case. Definitions by\ncourtesy of Christoph Lameter: who spent considerable effort on more scalable\nways of counting, but found insufficient benefit in practice.\n\nAnd adding an mm with swap to the mmlist for swapoff: the list is well-\nguarded by its own lock, but the list_empty check now has to be repeated\ninside it.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "4c21e2f2441dc5fbb957b030333f5a3f2d02dea7", "tree": "1f76d33bb1d76221c6424bc5fed080a4f91349a6", "parents": [ "b38c6845b695141259019e2b7c0fe6c32a6e720d" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:40 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:42 2005 -0700" }, "message": "[PATCH] mm: split page table lock\n\nChristoph Lameter demonstrated very poor scalability on the SGI 512-way, with\na many-threaded application which concurrently initializes different parts of\na large anonymous area.\n\nThis patch corrects that, by using a separate spinlock per page table page, to\nguard the page table entries in that page, instead of using the mm\u0027s single\npage_table_lock. (But even then, page_table_lock is still used to guard page\ntable allocation, and anon_vma allocation.)\n\nIn this implementation, the spinlock is tucked inside the struct page of the\npage table page: with a BUILD_BUG_ON in case it overflows - which it would in\nthe case of 32-bit PA-RISC with spinlock debugging enabled.\n\nSplitting the lock is not quite for free: another cacheline access. Ideally,\nI suppose we would use split ptlock only for multi-threaded processes on\nmulti-cpu machines; but deciding that dynamically would have its own costs.\nSo for now enable it by config, at some number of cpus - since the Kconfig\nlanguage doesn\u0027t support inequalities, let preprocessor compare that with\nNR_CPUS. But I don\u0027t think it\u0027s worth being user-configurable: for good\ntesting of both split and unsplit configs, split now at 4 cpus, and perhaps\nchange that to 8 later.\n\nThere is a benefit even for singly threaded processes: kswapd can be attacking\none part of the mm while another part is busy faulting.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "deceb6cd17e6dfafe4c4f81b1b4153bc41b2cb70", "tree": "2a722f50e8edef8609a49f65bfcb222e499c44cc", "parents": [ "c34d1b4d165c67b966bca4aba026443d7ff161eb" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:33 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:41 2005 -0700" }, "message": "[PATCH] mm: follow_page with inner ptlock\n\nFinal step in pushing down common core\u0027s page_table_lock. follow_page no\nlonger wants caller to hold page_table_lock, uses pte_offset_map_lock itself;\nand so no page_table_lock is taken in get_user_pages itself.\n\nBut get_user_pages (and get_futex_key) do then need follow_page to pin the\npage for them: take Daniel\u0027s suggestion of bitflags to follow_page.\n\nNeed one for WRITE, another for TOUCH (it was the accessed flag before:\nvanished along with check_user_page_readable, but surely get_numa_maps is\nwrong to mark every page it finds as accessed), another for GET.\n\nAnd another, ANON to dispose of untouched_anonymous_page: it seems silly for\nthat to descend a second time, let follow_page observe if there was no page\ntable and return ZERO_PAGE if so. Fix minor bug in that: check VM_LOCKED -\nmake_pages_present ought to make readonly anonymous present.\n\nGive get_numa_maps a cond_resched while we\u0027re there.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "c34d1b4d165c67b966bca4aba026443d7ff161eb", "tree": "27ffca9daba2a6b16d29bd508faf3e68bda2aad1", "parents": [ "c0718806cf955d5eb51ea77bffb5b21d9bba4972" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:32 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:41 2005 -0700" }, "message": "[PATCH] mm: kill check_user_page_readable\n\ncheck_user_page_readable is a problematic variant of follow_page. It\u0027s used\nonly by oprofile\u0027s i386 and arm backtrace code, at interrupt time, to\nestablish whether a userspace stackframe is currently readable.\n\nThis is problematic, because we want to push the page_table_lock down inside\nfollow_page, and later split it; whereas oprofile is doing a spin_trylock on\nit (in the i386 case, forgotten in the arm case), and needs that to pin\nperhaps two pages spanned by the stackframe (which might be covered by\ndifferent locks when we split).\n\nI think oprofile is going about this in the wrong way: it doesn\u0027t need to know\nthe area is readable (neither i386 nor arm uses read protection of user\npages), it doesn\u0027t need to pin the memory, it should simply\n__copy_from_user_inatomic, and see if that succeeds or not. Sorry, but I\u0027ve\nnot got around to devising the sparse __user annotations for this.\n\nThen we can eliminate check_user_page_readable, and return to a single\nfollow_page without the __follow_page variants.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "508034a32b819a2d40aa7ac0dbc8cd2e044c2de6", "tree": "906a8f0095af24f403b30d649d3ec1ffb4ff2f50", "parents": [ "8f4f8c164cb4af1432cc25eda82928ea4519ba72" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:30 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:41 2005 -0700" }, "message": "[PATCH] mm: unmap_vmas with inner ptlock\n\nRemove the page_table_lock from around the calls to unmap_vmas, and replace\nthe pte_offset_map in zap_pte_range by pte_offset_map_lock: all callers are\nnow safe to descend without page_table_lock.\n\nDon\u0027t attempt fancy locking for hugepages, just take page_table_lock in\nunmap_hugepage_range. Which makes zap_hugepage_range, and the hugetlb test in\nzap_page_range, redundant: unmap_vmas calls unmap_hugepage_range anyway. Nor\ndoes unmap_vmas have much use for its mm arg now.\n\nThe tlb_start_vma and tlb_end_vma in unmap_page_range are now called without\npage_table_lock: if they\u0027re implemented at all, they typically come down to\nflush_cache_range (usually done outside page_table_lock) and flush_tlb_range\n(which we already audited for the mprotect case).\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "8f4f8c164cb4af1432cc25eda82928ea4519ba72", "tree": "49cd3c62069df1f8d6c863b9806923de16c10e8b", "parents": [ "663b97f7efd001b0c56bd5fce059c5272725b86f" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:29 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:40 2005 -0700" }, "message": "[PATCH] mm: unlink vma before pagetables\n\nIn most places the descent from pgd to pud to pmd to pte holds mmap_sem\n(exclusively or not), which ensures that free_pgtables cannot be freeing page\ntables from any level at the same time. But truncation and reverse mapping\ndescend without mmap_sem.\n\nNo problem: just make sure that a vma is unlinked from its prio_tree (or\nnonlinear list) and from its anon_vma list, after zapping the vma, but before\nfreeing its page tables. Then neither vmtruncate nor rmap can reach that vma\nwhose page tables are now volatile (nor do they need to reach it, since all\nits page entries have been zapped by this stage).\n\nThe i_mmap_lock and anon_vma-\u003elock already serialize this correctly; but the\nlocking hierarchy is such that we cannot take them while holding\npage_table_lock. Well, we\u0027re trying to push that down anyway. So in this\npatch, move anon_vma_unlink and unlink_file_vma into free_pgtables, at the\nsame time as moving page_table_lock around calls to unmap_vmas.\n\ntlb_gather_mmu and tlb_finish_mmu then fall outside the page_table_lock, but\nwe made them preempt_disable and preempt_enable earlier; and a long source\naudit of all the architectures has shown no problem with removing\npage_table_lock from them. free_pgtables doesn\u0027t need page_table_lock for\nitself, nor for what it calls; tlb-\u003emm-\u003enr_ptes is usually protected by\npage_table_lock, but partly by non-exclusive mmap_sem - here it\u0027s decremented\nwith exclusive mmap_sem, or mm_users 0. update_hiwater_rss and\nvm_unacct_memory don\u0027t need page_table_lock either.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "8f4e2101fd7df9031a754eedb82e2060b51f8c45", "tree": "624db00c6160d70376a57447b45b935b293e396b", "parents": [ "b462705ac679f6195d1b23a752cda592d9107495" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:26 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:40 2005 -0700" }, "message": "[PATCH] mm: page fault handler locking\n\nOn the page fault path, the patch before last pushed acquiring the\npage_table_lock down to the head of handle_pte_fault (though it\u0027s also taken\nand dropped earlier when a new page table has to be allocated).\n\nNow delete that line, read \"entry \u003d *pte\" without it, and go off to this or\nthat page fault handler on the basis of this unlocked peek. Usually the\nhandler can proceed without the lock, relying on the subsequent locked\npte_same or pte_none test to back out when necessary; though do_wp_page needs\nthe lock immediately, and do_file_page doesn\u0027t check (if there\u0027s a race,\ninstall_page just zaps the entry and reinstalls it).\n\nBut on those architectures (notably i386 with PAE) whose pte is too big to be\nread atomically, if SMP or preemption is enabled, do_swap_page and\ndo_file_page might cause irretrievable damage if passed a Frankenstein entry\nstitched together from unrelated parts. In those configs, \"pte_unmap_same\"\nhas to take page_table_lock, validate orig_pte still the same, and drop\npage_table_lock before unmapping, before proceeding.\n\nUse pte_offset_map_lock and pte_unmap_unlock throughout the handlers; but lock\navoidance leaves more lone maps and unmaps than elsewhere.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "c74df32c724a1652ad8399b4891bb02c9d43743a", "tree": "5a79d56fdcf7dc2053a277dbf6db7c3b339e9659", "parents": [ "1bb3630e89cb8a7b3d3807629c20c5bad88290ff" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:23 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:40 2005 -0700" }, "message": "[PATCH] mm: ptd_alloc take ptlock\n\nSecond step in pushing down the page_table_lock. Remove the temporary\nbridging hack from __pud_alloc, __pmd_alloc, __pte_alloc: expect callers not\nto hold page_table_lock, whether it\u0027s on init_mm or a user mm; take\npage_table_lock internally to check if a racing task already allocated.\n\nConvert their callers from common code. But avoid coming back to change them\nagain later: instead of moving the spin_lock(\u0026mm-\u003epage_table_lock) down,\nswitch over to new macros pte_alloc_map_lock and pte_unmap_unlock, which\nencapsulate the mapping+locking and unlocking+unmapping together, and in the\nend may use alternatives to the mm page_table_lock itself.\n\nThese callers all hold mmap_sem (some exclusively, some not), so at no level\ncan a page table be whipped away from beneath them; and pte_alloc uses the\n\"atomic\" pmd_present to test whether it needs to allocate. It appears that on\nall arches we can safely descend without page_table_lock.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "1bb3630e89cb8a7b3d3807629c20c5bad88290ff", "tree": "3d1fd73487ca66f227701b9530f2c76fcc6f9da4", "parents": [ "872fec16d9a0ed3b75b8893aa217e49cca575ee5" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:22 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:40 2005 -0700" }, "message": "[PATCH] mm: ptd_alloc inline and out\n\nIt seems odd to me that, whereas pud_alloc and pmd_alloc test inline, only\ncalling out-of-line __pud_alloc __pmd_alloc if allocation needed,\npte_alloc_map and pte_alloc_kernel are entirely out-of-line. Though it does\nadd a little to kernel size, change them to macros testing inline, calling\n__pte_alloc or __pte_alloc_kernel to allocate out-of-line. Mark none of them\nas fastcalls, leave that to CONFIG_REGPARM or not.\n\nIt also seems more natural for the out-of-line functions to leave the offset\ncalculation and map to the inline, which has to do it anyway for the common\ncase. At least mremap move wants __pte_alloc without _map.\n\nMacros rather than inline functions, certainly to avoid the header file issues\nwhich arise from CONFIG_HIGHPTE needing kmap_types.h, but also in case any\narchitectures I haven\u0027t built would have other such problems.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "872fec16d9a0ed3b75b8893aa217e49cca575ee5", "tree": "1dfc8b9f2754bdfff645188e497865c00201d535", "parents": [ "46dea3d092d23a58b42499cc8a21de0fad079f4a" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:21 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:40 2005 -0700" }, "message": "[PATCH] mm: init_mm without ptlock\n\nFirst step in pushing down the page_table_lock. init_mm.page_table_lock has\nbeen used throughout the architectures (usually for ioremap): not to serialize\nkernel address space allocation (that\u0027s usually vmlist_lock), but because\npud_alloc,pmd_alloc,pte_alloc_kernel expect caller holds it.\n\nReverse that: don\u0027t lock or unlock init_mm.page_table_lock in any of the\narchitectures; instead rely on pud_alloc,pmd_alloc,pte_alloc_kernel to take\nand drop it when allocating a new one, to check lest a racing task already\ndid. Similarly no page_table_lock in vmalloc\u0027s map_vm_area.\n\nSome temporary ugliness in __pud_alloc and __pmd_alloc: since they also handle\nuser mms, which are converted only by a later patch, for now they have to lock\ndifferently according to whether or not it\u0027s init_mm.\n\nIf sources get muddled, there\u0027s a danger that an arch source taking\ninit_mm.page_table_lock will be mixed with common source also taking it (or\nneither take it). So break the rules and make another change, which should\nbreak the build for such a mismatch: remove the redundant mm arg from\npte_alloc_kernel (ppc64 scrapped its distinct ioremap_mm in 2.6.13).\n\nExceptions: arm26 used pte_alloc_kernel on user mm, now pte_alloc_map; ia64\nused pte_alloc_map on init_mm, now pte_alloc_kernel; parisc had bad args to\npmd_alloc and pte_alloc_kernel in unused USE_HPPA_IOREMAP code; ppc64\nmap_io_page forgot to unlock on failure; ppc mmu_mapin_ram and ppc64 im_free\ntook page_table_lock for no good reason.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "365e9c87a982c03d0af3886e29d877f581b59611", "tree": "d06c1918ca9fe6677d7e4e869555e095004274f7", "parents": [ "861f2fb8e796022b4928cab9c74fca6681a1c557" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:18 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:39 2005 -0700" }, "message": "[PATCH] mm: update_hiwaters just in time\n\nupdate_mem_hiwater has attracted various criticisms, in particular from those\nconcerned with mm scalability. Originally it was called whenever rss or\ntotal_vm got raised. Then many of those callsites were replaced by a timer\ntick call from account_system_time. Now Frank van Maarseveen reports that to\nbe found inadequate. How about this? Works for Frank.\n\nReplace update_mem_hiwater, a poor combination of two unrelated ops, by macros\nupdate_hiwater_rss and update_hiwater_vm. Don\u0027t attempt to keep\nmm-\u003ehiwater_rss up to date at timer tick, nor every time we raise rss (usually\nby 1): those are hot paths. Do the opposite, update only when about to lower\nrss (usually by many), or just before final accounting in do_exit. Handle\nmm-\u003ehiwater_vm in the same way, though it\u0027s much less of an issue. Demand\nthat whoever collects these hiwater statistics do the work of taking the\nmaximum with rss or total_vm.\n\nAnd there has been no collector of these hiwater statistics in the tree. The\nnew convention needs an example, so match Frank\u0027s usage by adding a VmPeak\nline above VmSize to /proc/\u003cpid\u003e/status, and also a VmHWM line above VmRSS\n(High-Water-Mark or High-Water-Memory).\n\nThere was a particular anomaly during mremap move, that hiwater_vm might be\ncaptured too high. A fleeting such anomaly remains, but it\u0027s quickly\ncorrected now, whereas before it would stick.\n\nWhat locking? None: if the app is racy then these statistics will be racy,\nit\u0027s not worth any overhead to make them exact. But whenever it suits,\nhiwater_vm is updated under exclusive mmap_sem, and hiwater_rss under\npage_table_lock (for now) or with preemption disabled (later on): without\ngoing to any trouble, minimize the time between reading current values and\nupdating, to minimize those occasions when a racing thread bumps a count up\nand back down in between.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "9e9bef07ce5a342aa6246ebc5c20829d0d5d63d0", "tree": "6f7286ebbb2c4b8af7bb04897a64b054c67faa4a", "parents": [ "86d912f41dca32eca8827f2f878139735e69dc28" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:15 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:39 2005 -0700" }, "message": "[PATCH] mm: do_swap_page race major\n\nSmall adjustment: do_swap_page should report its !pte_same race as a major\nfault if it had to read into swap cache, because whatever raced with it will\nhave found page already in cache and reported minor fault.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "86d912f41dca32eca8827f2f878139735e69dc28", "tree": "9f29b853e7690f0fa10c4257a2b8649673908e2d", "parents": [ "8c10376271e097fa13cda956e1b2f3cb7e4d4dd9" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:14 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:39 2005 -0700" }, "message": "[PATCH] mm: zap_pte_range dec rss\n\nSmall adjustment: zap_pte_range decrement its rss counts from 0 then finally\nadd, avoiding negations - we don\u0027t have or need a sub_mm_rss.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "8c10376271e097fa13cda956e1b2f3cb7e4d4dd9", "tree": "0bb1c428bfddce70eb1195c625add027899416e0", "parents": [ "b5810039a54e5babf428e9a1e89fc1940fabff11" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:13 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:39 2005 -0700" }, "message": "[PATCH] mm: copy_one_pte inc rss\n\nSmall adjustment, following Nick\u0027s suggestion: it\u0027s more straightforward for\ncopy_pte_range to let copy_one_pte do the rss incrementation, than use an\nindex it passed back. Saves a #define, and 16 bytes of .text.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "b5810039a54e5babf428e9a1e89fc1940fabff11", "tree": "835836cb527ec9bd525f93eb7e016f3dfb8c8ae2", "parents": [ "f9c98d0287de42221c624482fd4f8d485c98ab22" ], "author": { "name": "Nick Piggin", "email": "nickpiggin@yahoo.com.au", "time": "Sat Oct 29 18:16:12 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:39 2005 -0700" }, "message": "[PATCH] core remove PageReserved\n\nRemove PageReserved() calls from core code by tightening VM_RESERVED\nhandling in mm/ to cover PageReserved functionality.\n\nPageReserved special casing is removed from get_page and put_page.\n\nAll setting and clearing of PageReserved is retained, and it is now flagged\nin the page_alloc checks to help ensure we don\u0027t introduce any refcount\nbased freeing of Reserved pages.\n\nMAP_PRIVATE, PROT_WRITE of VM_RESERVED regions is tentatively being\ndeprecated. We never completely handled it correctly anyway, and is be\nreintroduced in future if required (Hugh has a proof of concept).\n\nOnce PageReserved() calls are removed from kernel/power/swsusp.c, and all\narch/ and driver code, the Set and Clear calls, and the PG_reserved bit can\nbe trivially removed.\n\nLast real user of PageReserved is swsusp, which uses PageReserved to\ndetermine whether a struct page points to valid memory or not. This still\nneeds to be addressed (a generic page_is_ram() should work).\n\nA last caveat: the ZERO_PAGE is now refcounted and managed with rmap (and\nthus mapcounted and count towards shared rss). These writes to the struct\npage could cause excessive cacheline bouncing on big systems. There are a\nnumber of ways this could be addressed if it is an issue.\n\nSigned-off-by: Nick Piggin \u003cnpiggin@suse.de\u003e\n\nRefcount bug fix for filemap_xip.c\n\nSigned-off-by: Carsten Otte \u003ccotte@de.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "ae859762332f19bfc06f4c4a1b1fefb41e9e1084", "tree": "4f21583bb1441e5555ed199a40d5f679bb4506e9", "parents": [ "4294621f41a85497019fae64341aa5351a1921b7" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:05 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:38 2005 -0700" }, "message": "[PATCH] mm: batch updating mm_counters\n\ntlb_finish_mmu used to batch zap_pte_range\u0027s update of mm rss, which may be\nworthwhile if the mm is contended, and would reduce atomic operations if the\ncounts were atomic. Let zap_pte_range now batch its updates to file_rss and\nanon_rss, per page-table in case we drop the lock outside; and copy_pte_range\nbatch them too.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "4294621f41a85497019fae64341aa5351a1921b7", "tree": "fdeb7eb44384a99d0679ffa6de5019bab0ea2166", "parents": [ "404351e67a9facb475abf1492245374a28d13e90" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:05 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:38 2005 -0700" }, "message": "[PATCH] mm: rss \u003d file_rss + anon_rss\n\nI was lazy when we added anon_rss, and chose to change as few places as\npossible. So currently each anonymous page has to be counted twice, in rss\nand in anon_rss. Which won\u0027t be so good if those are atomic counts in some\nconfigurations.\n\nChange that around: keep file_rss and anon_rss separately, and add them\ntogether (with get_mm_rss macro) when the total is needed - reading two\natomics is much cheaper than updating two atomics. And update anon_rss\nupfront, typically in memory.c, not tucked away in page_add_anon_rmap.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "fc2acab31be8e869b2d5f6de12f557f6f054f19c", "tree": "60cf419f5e88c3c46d39675a14649ea1e5849f03", "parents": [ "4d6ddfa9242bc3d27fb0f7248f6fdee0299c731f" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:03 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:37 2005 -0700" }, "message": "[PATCH] mm: tlb_finish_mmu forget rss\n\nzap_pte_range has been counting the pages it frees in tlb-\u003efreed, then\ntlb_finish_mmu has used that to update the mm\u0027s rss. That got stranger when I\nadded anon_rss, yet updated it by a different route; and stranger when rss and\nanon_rss became mm_counters with special access macros. And it would no\nlonger be viable if we\u0027re relying on page_table_lock to stabilize the\nmm_counter, but calling tlb_finish_mmu outside that lock.\n\nRemove the mmu_gather\u0027s freed field, let tlb_finish_mmu stick to its own\nbusiness, just decrement the rss mm_counter in zap_pte_range (yes, there was\nsome point to batching the update, and a subsequent patch restores that). And\nforget the anal paranoia of first reading the counter to avoid going negative\n- if rss does go negative, just fix that bug.\n\nRemove the mmu_gather\u0027s flushes and avoided_flushes from arm and arm26: no use\nwas being made of them. But arm26 alone was actually using the freed, in the\nway some others use need_flush: give it a need_flush. arm26 seems to prefer\nspaces to tabs here: respect that.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "4d6ddfa9242bc3d27fb0f7248f6fdee0299c731f", "tree": "da5b753df64e7163a35487005e50a3b90b0b0b9b", "parents": [ "15a23ffa2fc91cebdac44d4aee994f59d5c28dc0" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:16:02 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:37 2005 -0700" }, "message": "[PATCH] mm: tlb_is_full_mm was obscure\n\ntlb_is_full_mm? What does that mean? The TLB is full? No, it means that the\nmm\u0027s last user has gone and the whole mm is being torn down. And it\u0027s an\ninline function because sparc64 uses a different (slightly better)\n\"tlb_frozen\" name for the flag others call \"fullmm\".\n\nAnd now the ptep_get_and_clear_full macro used in zap_pte_range refers\ndirectly to tlb-\u003efullmm, which would be wrong for sparc64. Rather than\ncorrect that, I\u0027d prefer to scrap tlb_is_full_mm altogether, and change\nsparc64 to just use the same poor name as everyone else - is that okay?\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "65500d234e74fc4e8f18e1a429bc24e51e75de4a", "tree": "2bae8c3622b6537dbd142ba2744c7cc9430d3b69", "parents": [ "7c1fd6b964860cdcf44b6b98d7dcd8cc16a0a26d" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:15:59 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:37 2005 -0700" }, "message": "[PATCH] mm: page fault handlers tidyup\n\nImpose a little more consistency on the page fault handlers do_wp_page,\ndo_swap_page, do_anonymous_page, do_no_page, do_file_page: why not pass their\narguments in the same order, called the same names?\n\nbreak_cow is all very well, but what it did was inlined elsewhere: easier to\ncompare if it\u0027s brought back into do_wp_page.\n\ndo_file_page\u0027s fallback to do_no_page dates from a time when we were testing\npte_file by using it wherever possible: currently it\u0027s peculiar to nonlinear\nvmas, so just check that. BUG_ON if not? Better not, it\u0027s probably page\ntable corruption, so just show the pte: hmm, there\u0027s a pte_ERROR macro, let\u0027s\nuse that for do_wp_page\u0027s invalid pfn too.\n\nHah! Someone in the ppc64 world noticed pte_ERROR was unused so removed it:\nrestored (and say \"pud\" not \"pmd\" in its pud_ERROR).\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "72866f6f277ec0ddd6df7a3b6ecdcf59a28de115", "tree": "a76bb7fb8e092250c4399c20f0c7c973b77cb0bb", "parents": [ "6237bcd94851e9cf0ecd2520d744779df0f5a9a6" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:15:55 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:36 2005 -0700" }, "message": "[PATCH] mm: anon is already wrprotected\n\ndo_anonymous_page\u0027s pte_wrprotect causes some confusion: in such a case,\nvm_page_prot must already be forcing COW, so must omit write permission, and\nso the pte_wrprotect is redundant. Replace it by a comment to that effect,\nand reword the comment on unuse_pte which also caused confusion.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "6237bcd94851e9cf0ecd2520d744779df0f5a9a6", "tree": "5f34366cd527382489bd0bebbac8d39d59f1d5db", "parents": [ "0c942a4539c09adf09097315cc174aefd0eeedf7" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:15:54 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:36 2005 -0700" }, "message": "[PATCH] mm: zap_pte_range dont dirty anon\n\nzap_pte_range already avoids wasting time to mark_page_accessed on anon pages:\nit can also skip anon set_page_dirty - the page only needs to be marked dirty\nif shared with another mm, but that will say pte_dirty too.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "e040f218bb49a6965a5b77edce05fe47a62dda39", "tree": "e5547e04fa312f1b87db66e4ac2b6eed885ca276", "parents": [ "09ad4bbc3a5c93316d7f4ffc0c310d9cbb28c2f0" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Oct 29 18:15:53 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Oct 29 21:40:36 2005 -0700" }, "message": "[PATCH] mm: copy_pte_range progress fix\n\nMy latency breaking in copy_pte_range didn\u0027t work as intended: instead of\nchecking at regularish intervals, after the first interval it checked every\ntime around the loop, too impatient to be preempted. Fix that.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "ac9b9c667c2e1194e22ebe0a441ae1c37aaa9b90", "tree": "3903d87d0b56a49ead39c0460b5bc0b86b040775", "parents": [ "93918e9afc76717176e9e114e79cdbb602a45ae8" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Thu Oct 20 16:24:28 2005 +0100" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Thu Oct 20 09:02:07 2005 -0700" }, "message": "[PATCH] Fix handling spurious page fault for hugetlb region\n\nThis reverts commit 3359b54c8c07338f3a863d1109b42eebccdcf379 and\nreplaces it with a cleaner version that is purely based on page table\noperations, so that the synchronization between inode size and hugetlb\nmappings becomes moot.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "3359b54c8c07338f3a863d1109b42eebccdcf379", "tree": "f91edd52c71e57ce4b46e3875c9054666ca4e24c", "parents": [ "bb7e257ef8d8ba43cab356aa1cc1b20d0106d45f" ], "author": { "name": "Seth, Rohit", "email": "rohit.seth@intel.com", "time": "Tue Oct 18 14:15:12 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Wed Oct 19 13:56:27 2005 -0700" }, "message": "[PATCH] Handle spurious page fault for hugetlb region\n\nThe hugetlb pages are currently pre-faulted. At the time of mmap of\nhugepages, we populate the new PTEs. It is possible that HW has already\ncached some of the unused PTEs internally. These stale entries never\nget a chance to be purged in existing control flow.\n\nThis patch extends the check in page fault code for hugepages. Check if\na faulted address falls with in size for the hugetlb file backing it.\nWe return VM_FAULT_MINOR for these cases (assuming that the arch\nspecific page-faulting code purges the stale entry for the archs that\nneed it).\n\nSigned-off-by: Rohit Seth \u003crohit.seth@intel.com\u003e\n\n[ This is apparently arguably an ia64 port bug. But the code won\u0027t\n hurt, and for now it fixes a real problem on some ia64 machines ]\n\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "5ce7852cdf07ab903fb1c72d0915ac492c6e07c7", "tree": "7992224cd306be5e827de0bdf6255bc87a2bf4c8", "parents": [ "dd81eca83c8300c95d8a1eaf0d38f56513711535" ], "author": { "name": "Adrian Bunk", "email": "bunk@stusta.de", "time": "Sat Sep 10 00:26:28 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Sat Sep 10 10:06:25 2005 -0700" }, "message": "[PATCH] mm/filemap.c: make two functions static\n\nWith Nick Piggin \u003cnpiggin@suse.de\u003e\n\nGive some things static scope.\n\nSigned-off-by: Adrian Bunk \u003cbunk@stusta.de\u003e\nSigned-off-by: Nick Piggin \u003cnpiggin@suse.de\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "a600388d28419305aad3c4c0af52c223cf6fa0af", "tree": "c70d3d80275f189c49311183472367f45d1a1ef2", "parents": [ "fa5b08d5f818063d18433194f20359ef2ae50254" ], "author": { "name": "Zachary Amsden", "email": "zach@vmware.com", "time": "Sat Sep 03 15:55:04 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@evo.osdl.org", "time": "Mon Sep 05 00:05:48 2005 -0700" }, "message": "[PATCH] x86: ptep_clear optimization\n\nAdd a new accessor for PTEs, which passes the full hint from the mmu_gather\nstruct; this allows architectures with hardware pagetables to optimize away\natomic PTE operations when destroying an address space. Removing the\nlocked operation should allow better pipelining of memory access in this\nloop. I measured an average savings of 30-35 cycles per zap_pte_range on\nthe first 500 destructions on Pentium-M, but I believe the optimization\nwould win more on older processors which still assert the bus lock on xchg\nfor an exclusive cacheline.\n\nUpdate: I made some new measurements, and this saves exactly 26 cycles over\nptep_get_and_clear on Pentium M. On P4, with a PAE kernel, this saves 180\ncycles per ptep_get_and_clear, for a whopping 92160 cycles savings for a\nfull address space destruction.\n\npte_clear_full is not yet used, but is provided for future optimizations\n(in particular, when running inside of a hypervisor that queues page table\nupdates, the full hint allows us to avoid queueing unnecessary page table\nupdate for an address space in the process of being destroyed.\n\nThis is not a huge win, but it does help a bit, and sets the stage for\nfurther hypervisor optimization of the mm layer on all architectures.\n\nSigned-off-by: Zachary Amsden \u003czach@vmware.com\u003e\nCc: Christoph Lameter \u003cchristoph@lameter.com\u003e\nCc: \u003clinux-mm@kvack.org\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "4944e76d81801b8e60ed3e7789443f210c16ed65", "tree": "d20096acf35b1af03eba1d043e6376a154cae054", "parents": [ "d44ed4f86892e350f4b16a3489b7e7c1a9bb7ead" ], "author": { "name": "Paolo \u0027Blaisorblade\u0027 Giarrusso", "email": "blaisorblade@yahoo.it", "time": "Sat Sep 03 15:54:56 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@evo.osdl.org", "time": "Mon Sep 05 00:05:45 2005 -0700" }, "message": "[PATCH] mm: remove implied vm_ops check\n\nIf !vma-\u003evm-ops we already BUG above, so retesting it is useless. The\ncompiler cannot optimize this because BUG is a macro and is not thus marked\nnoreturn; that should possibly be fixed.\n\nSigned-off-by: Paolo \u0027Blaisorblade\u0027 Giarrusso \u003cblaisorblade@yahoo.it\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "d992895ba2b27cf5adf1ba0ad6d27662adc54c5e", "tree": "65a4d1f18a93a9e89d43fe0b8e0b3009675c50f0", "parents": [ "40193713df2cdb9c233b3fc2029ecdccb40cb1e4" ], "author": { "name": "Nick Piggin", "email": "nickpiggin@yahoo.com.au", "time": "Sun Aug 28 16:49:11 2005 +1000" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Mon Aug 29 17:25:04 2005 -0700" }, "message": "[PATCH] Lazy page table copies in fork()\n\nDefer copying of ptes until fault time when it is possible to reconstruct\nthe pte from backing store. Idea from Andi Kleen and Nick Piggin.\n\nThanks to input from Rik van Riel and Linus and to Hugh for correcting\nmy blundering.\n\nRay Fucillo \u003cfucillo@intersystems.com\u003e reports:\n\n \"I applied this latest patch to a 2.6.12 kernel and found that it does\n resolve the problem. Prior to the patch on this machine, I was\n seeing about 23ms spent in fork for ever 100MB of shared memory\n segment.\n\n After applying the patch, fork is taking about 1ms regardless of the\n shared memory size.\"\n\nSigned-off-by: Nick Piggin \u003cnpiggin@suse.de\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "a68d2ebc1581a3aec57bd032651e013fa609f530", "tree": "b41977c7157d7e26f37e9cb502cd1afbbddcbc17", "parents": [ "f33ea7f404e592e4563b12101b7a4d17da6558d7" ], "author": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Wed Aug 03 10:07:09 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Wed Aug 03 10:07:09 2005 -0700" }, "message": "Fix up recent get_user_pages() handling\n\nThe VM_FAULT_WRITE thing is an extra bit, not a valid return value, and\nhas to be treated as such by get_user_pages().\n\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "f33ea7f404e592e4563b12101b7a4d17da6558d7", "tree": "1d587ad8a06cb6d2e3a187f0312c8a524ffefe53", "parents": [ "5cb4cc0d8211c490537c8568001958fc76741312" ], "author": { "name": "Nick Piggin", "email": "nickpiggin@yahoo.com.au", "time": "Wed Aug 03 20:24:01 2005 +1000" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Wed Aug 03 09:12:05 2005 -0700" }, "message": "[PATCH] fix get_user_pages bug\n\nChecking pte_dirty instead of pte_write in __follow_page is problematic\nfor s390, and for copy_one_pte which leaves dirty when clearing write.\n\nSo revert __follow_page to check pte_write as before, and make\ndo_wp_page pass back a special extra VM_FAULT_WRITE bit to say it has\ndone its full job: once get_user_pages receives this value, it no longer\nrequires pte_write in __follow_page.\n\nBut most callers of handle_mm_fault, in the various architectures, have\nswitch statements which do not expect this new case. To avoid changing\nthem all in a hurry, make an inline wrapper function (using the old\nname) that masks off the new bit, and use the extended interface with\ndouble underscores.\n\nYes, we do have a call to do_wp_page from do_swap_page, but no need to\nchange that: in rare case it\u0027s needed, another do_wp_page will follow.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\n[ Cleanups by Nick Piggin ]\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "690dbe1ced143876d8fa56b72310738dbe079d0a", "tree": "4297d5647ce5e6cbe429dc506007579952c31015", "parents": [ "74f9c9c258249fba3e2e78f70691528426a6c010" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Mon Aug 01 21:11:42 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Mon Aug 01 21:38:00 2005 -0700" }, "message": "[PATCH] x86_64: access of some bad address\n\nx86_64 has a large sparse gate area between VSYSCALL_START and\nVSYSCALL_END, not all of it presently backed by pmds. Alexander Nyberg has\nfound that in some circumstances gdb may try to ptrace here, and hit\nget_user_pages BUG_ON. It seems odd that gdb should be accessing here, but\nit certainly shouldn\u0027t crash in this way: relax BUG_ON to -EFAULT. Fixes\nkernel bugzilla #4801.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nCc: Andi Kleen \u003cak@suse.de\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "4ceb5db9757aaeadcf8fbbf97d76bd42aa4df0d6", "tree": "6a3108ceea457c21130838d49736f5e9de3badc3", "parents": [ "8d894c47975f7222c5537e450e71310b395488c7" ], "author": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Mon Aug 01 11:14:49 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Mon Aug 01 11:14:49 2005 -0700" }, "message": "Fix get_user_pages() race for write access\n\nThere\u0027s no real guarantee that handle_mm_fault() will always be able to\nbreak a COW situation - if an update from another thread ends up\nmodifying the page table some way, handle_mm_fault() may end up\nrequiring us to re-try the operation.\n\nThat\u0027s normally fine, but get_user_pages() ended up re-trying it as a\nread, and thus a write access could in theory end up losing the dirty\nbit or be done on a page that had not been properly COW\u0027ed.\n\nThis makes get_user_pages() always retry write accesses as write\naccesses by making \"follow_page()\" require that a writable follow has\nthe dirty bit set. That simplifies the code and solves the race: if the\nCOW break fails for some reason, we\u0027ll just loop around and try again.\n\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "1aaf18ff9de1f37bf674236fc0779c3aaa65b998", "tree": "53df4d3dd0fbd81b79d5cdb63cf0d11853307a6b", "parents": [ "0cfc11ed45e4c00750039e5a18c0fc0d681e19db" ], "author": { "name": "Andrew Morton", "email": "akpm@osdl.org", "time": "Wed Jul 27 11:43:54 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@g5.osdl.org", "time": "Wed Jul 27 16:25:53 2005 -0700" }, "message": "[PATCH] check_user_page_readable() deadlock fix\n\nFix bug identifued by Richard Purdie \u003crpurdie@rpsys.net\u003e.\n\noprofile calls check_user_page_readable() from interrupt context, so we\ndeadlock over various VFS locks.\n\nBut check_user_page_readable() doesn\u0027t imply either a read or a write of the\npage\u0027s contents. Change __follow_page() so that check_user_page_readable()\ncan tell __follow_page() that we\u0027re not accessing the page\u0027s contents, and use\nthat info to avoid the troublesome lock-takings.\n\nAlso, make follow_page() inline for the single callsite in memory.c to save a\nbit of stack space.\n\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "2d15cab85b85a56cc886037cab43cc292923ff22", "tree": "55de7526dfb766067821d552892ad43b32f78c35", "parents": [ "1e8a81c5a37907bc082025d3468718116dca1eeb" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Sat Jun 25 14:54:33 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Sat Jun 25 16:24:26 2005 -0700" }, "message": "[PATCH] mm: fix remap_pte_range BUG\n\nOut-of-tree user of remap_pfn_range hit kernel BUG at mm/memory.c:1112! It\npasses an unrounded size to remap_pfn_range, which was okay before 2.6.12,\nbut misses remap_pte_range\u0027s new end condition. An audit of all the other\nptwalks confirms that this is the only one so exposed.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "3d41088fa327782b14b5659dbcfff62ec704c23c", "tree": "bd9a25f9ae7a0e232c9d5006849905e31a9dbc92", "parents": [ "363412b4f70a2ba19c76a01da7580472399312d4" ], "author": { "name": "Martin Waitz", "email": "tali@admingilde.org", "time": "Thu Jun 23 22:05:21 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Fri Jun 24 00:06:40 2005 -0700" }, "message": "[PATCH] DocBook: update comments\n\nThis patch updates some comments to match code changes.\n\nSigned-off-by: Martin Waitz \u003ctali@admingilde.org\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "d41dee369bff3b9dcb6328d4d822926c28cc2594", "tree": "a0405f3b7af3ebca21838a7d427bd75a067bf850", "parents": [ "af705362ab6018071310c5fcd436a6b457517d5f" ], "author": { "name": "Andy Whitcroft", "email": "apw@shadowen.org", "time": "Thu Jun 23 00:07:54 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Thu Jun 23 09:45:04 2005 -0700" }, "message": "[PATCH] sparsemem memory model\n\nSparsemem abstracts the use of discontiguous mem_maps[]. This kind of\nmem_map[] is needed by discontiguous memory machines (like in the old\nCONFIG_DISCONTIGMEM case) as well as memory hotplug systems. Sparsemem\nreplaces DISCONTIGMEM when enabled, and it is hoped that it can eventually\nbecome a complete replacement.\n\nA significant advantage over DISCONTIGMEM is that it\u0027s completely separated\nfrom CONFIG_NUMA. When producing this patch, it became apparent in that NUMA\nand DISCONTIG are often confused.\n\nAnother advantage is that sparse doesn\u0027t require each NUMA node\u0027s ranges to be\ncontiguous. It can handle overlapping ranges between nodes with no problems,\nwhere DISCONTIGMEM currently throws away that memory.\n\nSparsemem uses an array to provide different pfn_to_page() translations for\neach SECTION_SIZE area of physical memory. This is what allows the mem_map[]\nto be chopped up.\n\nIn order to do quick pfn_to_page() operations, the section number of the page\nis encoded in page-\u003eflags. Part of the sparsemem infrastructure enables\nsharing of these bits more dynamically (at compile-time) between the\npage_zone() and sparsemem operations. However, on 32-bit architectures, the\nnumber of bits is quite limited, and may require growing the size of the\npage-\u003eflags type in certain conditions. Several things might force this to\noccur: a decrease in the SECTION_SIZE (if you want to hotplug smaller areas of\nmemory), an increase in the physical address space, or an increase in the\nnumber of used page-\u003eflags.\n\nOne thing to note is that, once sparsemem is present, the NUMA node\ninformation no longer needs to be stored in the page-\u003eflags. It might provide\nspeed increases on certain platforms and will be stored there if there is\nroom. But, if out of room, an alternate (theoretically slower) mechanism is\nused.\n\nThis patch introduces CONFIG_FLATMEM. It is used in almost all cases where\nthere used to be an #ifndef DISCONTIG, because SPARSEMEM and DISCONTIGMEM\noften have to compile out the same areas of code.\n\nSigned-off-by: Andy Whitcroft \u003capw@shadowen.org\u003e\nSigned-off-by: Dave Hansen \u003chaveblue@us.ibm.com\u003e\nSigned-off-by: Martin Bligh \u003cmbligh@aracnet.com\u003e\nSigned-off-by: Adrian Bunk \u003cbunk@stusta.de\u003e\nSigned-off-by: Yasunori Goto \u003cy-goto@jp.fujitsu.com\u003e\nSigned-off-by: Bob Picco \u003cbob.picco@hp.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "c475a8ab625d567eacf5e30ec35d6d8704558062", "tree": "0971bef7b876f1b3eb160621fc2b61cb5313827b", "parents": [ "d296e9cd02c92e576ecce5344026a4df4353cdb2" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Tue Jun 21 17:15:12 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Tue Jun 21 18:46:21 2005 -0700" }, "message": "[PATCH] can_share_swap_page: use page_mapcount\n\nRemember that ironic get_user_pages race? when the raised page_count on a\npage swapped out led do_wp_page to decide that it had to copy on write, so\nsubstituted a different page into userspace. 2.6.7 onwards have Andrea\u0027s\nsolution, where try_to_unmap_one backs out if it finds page_count raised.\n\nWhich works, but is unsatisfying (rmap.c has no other page_count heuristics),\nand was found a few months ago to hang an intensive page migration test. A\nyear ago I was hesitant to engage page_mapcount, now it seems the right fix.\n\nSo remove the page_count hack from try_to_unmap_one; and use activate_page in\nunuse_mm when dropping lock, to replace its secondary effect of helping\nswapoff to make progress in that case.\n\nSimplify can_share_swap_page (now called only on anonymous pages) to check\npage_mapcount + page_swapcount \u003d\u003d 1: still needs the page lock to stabilize\ntheir (pessimistic) sum, but does not need swapper_space.tree_lock for that.\n\nIn do_swap_page, move swap_free and unlock_page below page_add_anon_rmap, to\nkeep sum on the high side, and correct when can_share_swap_page called.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "d296e9cd02c92e576ecce5344026a4df4353cdb2", "tree": "76af62c5d2c16e89672f71f2f4c7a65aa36ff914", "parents": [ "08ef472937e918875a82fd350d3de138aac50414" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Tue Jun 21 17:15:11 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Tue Jun 21 18:46:21 2005 -0700" }, "message": "[PATCH] do_wp_page: cannot share file page\n\nA small optimization to do_wp_page\u0027s check for whether to avoid copy by\nreusing the page already mapped. It can never share a cached file page,\nnor can it share a reserved page (often the empty zero page), so it\u0027s a\nwaste of time to lock and unlock in those cases. Which nowadays can both\nbe neatly excluded by a preliminary PageAnon test.\n\nChristoph has reported that a preliminary page_count test proved valuable\nfor scalability here, but PageAnon covers more common cases all at once.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "08ef472937e918875a82fd350d3de138aac50414", "tree": "990a503a711b82f94ad78ea5794a46992db05228", "parents": [ "7c2f3fda5666c280bcd00ac3b86963270b23e796" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Tue Jun 21 17:15:10 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Tue Jun 21 18:46:21 2005 -0700" }, "message": "[PATCH] get_user_pages: kill get_page_map\n\nSince its birth, get_user_pages has been calling a misguided get_page_map\nfunction. follow_page has already returned NULL if the pfn is invalid, we\ncannot reach an invalid pfn from a validated struct page.\n\nRemove get_page_map, and the messy rewind in get_user_pages to cope with\nits failure. Oh, and could we please call that \"struct page *page\" like\neverywhere else, instead of \"struct page *map\"?\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "b81074800b98ac50b64d4c8d34e8abf0fda5e3d1", "tree": "01b5114255fcf1ea8ea15fabe08cc9782131e684", "parents": [ "c64610ba585fabb36be78782868277f3d9741a2e" ], "author": { "name": "Kirill Korotaev", "email": "dev@sw.ru", "time": "Mon May 16 21:53:50 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Tue May 17 07:59:20 2005 -0700" }, "message": "[PATCH] do_swap_page() can map random data if swap read fails\n\nThere is a bug in do_swap_page(): when swap page happens to be unreadable,\npage filled with random data is mapped into user address space. The fix is\nto check for PageUptodate and send SIGBUS in case of error.\n\nSigned-Off-By: Kirill Korotaev \u003cdev@sw.ru\u003e\nSigned-Off-By: Alexey Kuznetsov \u003ckuznet@ms2.inr.ac.ru\u003e\nAcked-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "3bf5ee95648c694bac4d13529563c230cd4fe5f2", "tree": "9430e6e4f4c3d586ecb7375cd780fd17694888c7", "parents": [ "ee39b37b23da0b6ec53a8ebe90ff41c016f8ae27" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Tue Apr 19 13:29:16 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org.(none)", "time": "Tue Apr 19 13:29:16 2005 -0700" }, "message": "[PATCH] freepgt: hugetlb_free_pgd_range\n\nia64 and ppc64 had hugetlb_free_pgtables functions which were no longer being\ncalled, and it wasn\u0027t obvious what to do about them.\n\nThe ppc64 case turns out to be easy: the associated tables are noted elsewhere\nand freed later, safe to either skip its hugetlb areas or go through the\nmotions of freeing nothing. Since ia64 does need a special case, restore to\nppc64 the special case of skipping them.\n\nThe ia64 hugetlb case has been broken since pgd_addr_end went in, though it\nprobably appeared to work okay if you just had one such area; in fact it\u0027s\nbeen broken much longer if you consider a long munmap spanning from another\nregion into the hugetlb region.\n\nIn the ia64 hugetlb region, more virtual address bits are available than in\nthe other regions, yet the page tables are structured the same way: the page\nat the bottom is larger. Here we need to scale down each addr before passing\nit to the standard free_pgd_range. Was about to write a hugely_scaled_down\nmacro, but found htlbpage_to_page already exists for just this purpose. Fixed\noff-by-one in ia64 is_hugepage_only_range.\n\nUninline free_pgd_range to make it available to ia64. Make sure the\nvma-gathering loop in free_pgtables cannot join a hugepage_only_range to any\nother (safe to join huges? probably but don\u0027t bother).\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "ee39b37b23da0b6ec53a8ebe90ff41c016f8ae27", "tree": "4af606913ab8f95551623b788c0c66c1f5902229", "parents": [ "e0da382c92626ad1d7f4b7527d19b80104d67a83" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Tue Apr 19 13:29:15 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org.(none)", "time": "Tue Apr 19 13:29:15 2005 -0700" }, "message": "[PATCH] freepgt: remove MM_VM_SIZE(mm)\n\nThere\u0027s only one usage of MM_VM_SIZE(mm) left, and it\u0027s a troublesome macro\nbecause mm doesn\u0027t contain the (32-bit emulation?) info needed. But it too is\nonly needed because we ignore the end from the vma list.\n\nWe could make flush_pgtables return that end, or unmap_vmas. Choose the\nlatter, since it\u0027s a natural fit with unmap_mapping_range_vma needing to know\nits restart addr. This does make more than minimal change, but if unmap_vmas\nhad returned the end before, this is how we\u0027d have done it, rather than\nstoring the break_addr in zap_details.\n\nunmap_vmas used to return count of vmas scanned, but that\u0027s just debug which\nhasn\u0027t been useful in a while; and if we want the map_count 0 on exit check\nback, it can easily come from the final remove_vm_struct loop.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "e0da382c92626ad1d7f4b7527d19b80104d67a83", "tree": "b3f455518c286ee14cb2755ced8808487bca7911", "parents": [ "9f6c6fc505560465be0964eb4da1b6ca97bd3951" ], "author": { "name": "Hugh Dickins", "email": "hugh@veritas.com", "time": "Tue Apr 19 13:29:15 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org.(none)", "time": "Tue Apr 19 13:29:15 2005 -0700" }, "message": "[PATCH] freepgt: free_pgtables use vma list\n\nRecent woes with some arches needing their own pgd_addr_end macro; and 4-level\nclear_page_range regression since 2.6.10\u0027s clear_page_tables; and its\nlong-standing well-known inefficiency in searching throughout the higher-level\npage tables for those few entries to clear and free: all can be blamed on\nignoring the list of vmas when we free page tables.\n\nReplace exit_mmap\u0027s clear_page_range of the total user address space by\nfree_pgtables operating on the mm\u0027s vma list; unmap_region use it in the same\nway, giving floor and ceiling beyond which it may not free tables. This\nbrings lmbench fork/exec/sh numbers back to 2.6.10 (unless preempt is enabled,\nin which case latency fixes spoil unmap_vmas throughput).\n\nBeware: the do_mmap_pgoff driver failure case must now use unmap_region\ninstead of zap_page_range, since a page table might have been allocated, and\ncan only be freed while it is touched by some vma.\n\nMove free_pgtables from mmap.c to memory.c, where its lower levels are adapted\nfrom the clear_page_range levels. (Most of free_pgtables\u0027 old code was\nactually for a non-existent case, prev not properly set up, dating from before\nhch gave us split_vma.) Pass mmu_gather** in the public interfaces, since we\nmight want to add latency lockdrops later; but no attempt to do so yet, going\nby vma should itself reduce latency.\n\nBut what if is_hugepage_only_range? Those ia64 and ppc64 cases need careful\nexamination: put that off until a later patch of the series.\n\nWhat of x86_64\u0027s 32bit vdso page __map_syscall32 maps outside any vma?\n\nAnd the range to sparc64\u0027s flush_tlb_pgtables? It\u0027s less clear to me now that\nwe need to do more than is done here - every PMD_SIZE ever occupied will be\nflushed, do we really have to flush every PGDIR_SIZE ever partially occupied? \nA shame to complicate it unnecessarily.\n\nSpecial thanks to David Miller for time spent repairing my ceilings.\n\nSigned-off-by: Hugh Dickins \u003chugh@veritas.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n" }, { "commit": "1da177e4c3f41524e886b7f1b8a0c1fc7321cac2", "tree": "0bba044c4ce775e45a88a51686b5d9f90697ea9d", "parents": [], "author": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Sat Apr 16 15:20:36 2005 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@ppc970.osdl.org", "time": "Sat Apr 16 15:20:36 2005 -0700" }, "message": "Linux-2.6.12-rc2\n\nInitial git repository build. I\u0027m not bothering with the full history,\neven though we have it. We can create a separate \"historical\" git\narchive of that later if we want to, and in the meantime it\u0027s about\n3.2GB when imported into git - space that would just make the early\ngit days unnecessarily complicated, when we don\u0027t have a lot of good\ninfrastructure for it.\n\nLet it rip!\n" } ] }