| 1 | ## Note: this is my personal todo and ideas list (alexh@) |
| 2 | |
| 3 | * add a non-persistent unionfs VFS |
| 4 | - while this does not fully replace unionfs, it would deal with a few of the situations where unionfs is useful |
| 5 | - additionally the complexity is much lower, since everything can be kept in memory |
| 6 | |
| 7 | * bring the samba3 hammer shadow copy foo to maturity |
| 8 | |
| 9 | * add a communication channel mechanism to dm |
| 10 | - essentially a way to send messages and receive responses to dm target instances |
| 11 | |
| 12 | * add another optimized bcopy/memcpy/memset/bzero version, using the AVX instructions |
| 13 | |
| 14 | * Add TrueCrypt support |
| 15 | - Simply add a userland tool that sets dm_target_crypt up with the right parameters, extracted from the TrueCrypt header |
| 16 | - https://github.com/bwalex/tc-play <- done! |
| 17 | - need passphrase read timeout support |
| 18 | - need unmap support |
| 19 | - cryptdisks: needs generic crypto options; multiple keyfile support |
| 20 | |
| 21 | * Update cryptsetup |
| 22 | |
| 23 | * Keep opencrypto up-to-date |
| 24 | - http://www.openbsd.org/cgi-bin/cvsweb/src/sys/crypto/ |
| 25 | |
| 26 | * Port hwpmc & dig into (boot-up) performance |
| 27 | |
| 28 | * linuxulator |
| 29 | - port to x86_64 |
| 30 | - separate out common arch parts (linprocfs, for example) |
| 31 | |
| 32 | * Fix the crash analysis script (or rather the programs it calls [some segfault]) |
| 33 | |
| 34 | * Take a look at updating lvm/dm/libdevicemapper |
| 35 | |
| 36 | * Take a look at importing libdm from NetBSD |
| 37 | - would allow tc-play and similarly newly developed foo to avoid GPL license issues |
| 38 | |
| 39 | * rip out the disk partitioning from the disk subsystem and implement it in a more general fashion |
| 40 | - crazy idea: as dm targets with an auto-configuration option! |
| 41 | - would require to be able to create dm targets with an arbitrary name and not in /dev/mapper |
| 42 | |
| 43 | * ATA (automatic) spindown (see FreeBSD current) |
| 44 | |
| 45 | * Update callout |
| 46 | http://svn.freebsd.org/viewvc/base?view=revision&revision=127969 |
| 47 | |
| 48 | * inv ctxsw rusage |
| 49 | - see irc logs |
| 50 | - some incorrect accounting going on, don't remember details :) |
| 51 | |
| 52 | * unionfs update |
| 53 | - make it work without whiteout |
| 54 | |
| 55 | |
| 56 | |
| 57 | ###Boring: |
| 58 | |
| 59 | * Fix ipsec, get rid of old ipsec |
| 60 | |
| 61 | * sync up vr |
| 62 | - Added VT6105M specific register definitions. VT6105M has the following hardware capabilities. |
| 63 | - Tx/Rx IP/TCP/UDP checksum offload. |
| 64 | - VLAN hardware tag insertion/extraction. Due to lack of information |
| 65 | for getting extracted VLAN tag in Rx path, VLAN hardware support |
| 66 | was not implemented yet. |
| 67 | - CAM(Content Addressable Memory) based 32 entry perfect multicast/ |
| 68 | VLAN filtering. |
| 69 | - 8 priority queues. |
| 70 | o Implemented CAM based 32 entry perfect multicast filtering for |
| 71 | VT6105M. If number of multicast entry is greater than 32, vr(4) |
| 72 | uses traditional hash based filtering. |
| 73 | |
| 74 | * RedZone, a buffer corruption protection for the kernel malloc(9) facility has been implemented. |
| 75 | - This detects both buffer underflows and overflows at runtime on free(9) and realloc(9), |
| 76 | and prints backtraces from where memory was allocated and from where it was freed. |
| 77 | - see irc log below. |
| 78 | |
| 79 | * port uart driver (?) |
| 80 | |
| 81 | * port wscons (?) or update syscons |
| 82 | - probably way too much effort (wscons) |
| 83 | |
| 84 | |
| 85 | * port usb4bsd |
| 86 | - wrapper is included for userland; should be easy to port |
| 87 | - http://svn.freebsd.org/viewvc/base?view=revision&revision=184610 |
| 88 | - http://turbocat.net/~hselasky/usb4bsd/ |
| 89 | - http://gitweb.dragonflybsd.org/~polachok/dragonfly.git/shortlog/refs/heads/usb2 |
| 90 | |
| 91 | * suspend/resume for SMP x86 |
| 92 | - http://lists.freebsd.org/pipermail/freebsd-acpi/2008-May/004879.html |
| 93 | |
| 94 | * AMD64 suspend/resume |
| 95 | - http://svn.freebsd.org/viewvc/base?view=revision&revision=189903 |
| 96 | |
| 97 | |
| 98 | * text dumps |
| 99 | |
| 100 | |
| 101 | |
| 102 | [alexh@leaf:~/home] $ roundup-server -p 8080 bt=bugtracker |
| 103 | |
| 104 | <pre> |
| 105 | -05:48- : dillon@: no, double frees to the object cache are nasty. It can't detect them. the object |
| 106 | winds up in the magazine array twice |
| 107 | -05:48- : dillon@: (and possibly different magazines, too) |
| 108 | -05:49- : alexh@: can't I just write some magic to a free object on the first objcache_put and check |
| 109 | if it's there on objcache_put? |
| 110 | -05:49- : alexh@: and clear it on objcache_get, anyways |
| 111 | -05:50- : dillon@: no, because the object is still may have live-initialized fields |
| 112 | -05:50- : dillon@: because it hasn't been dtor'ed yet (one of the features of the objcache, to avoid |
| 113 | having to reinitialize objects every time) |
| 114 | -05:50- : dillon@: the mbuf code uses that feature I think, probably other bits too |
| 115 | -05:51- : dillon@: theoretically we could allocate slightly larger objects and store a magic number at |
| 116 | offset [-1] or something like that, but it gets a little iffy doing that |
| 117 | -05:52- : dillon@: the objcache with the objcache malloc default could probably do something like that |
| 118 | I guess. |
| 119 | -05:52- : dillon@: I don't consider memory tracking to be a huge issue w/ dragonfly, though I like the |
| 120 | idea of being able to do it. It is a much bigger problem in FreeBSD due to the |
| 121 | large number of committers |
| 122 | |
| 123 | |
| 124 | -05:55- : dillon@: For the slab allocator you may be able to do something using the Zone header. |
| 125 | -05:55- : dillon@: the slab allocator in fact I think already has optional code to allocate a tracking |
| 126 | bitmap to detect double-frees |
| 127 | -05:56- : dillon@: sorry, I just remembered the bit about the power-of-2 allocations |
| 128 | -05:56- : dillon@: for example, power-of-2-sized allocations are guaranteed not only to be aligned on |
| 129 | that particular size boundary, but also to not cross a PAGE_BOUNDARY (unless the |
| 130 | size is > PAGE_SIZE) |
| 131 | -05:57- : dillon@: various subsystems such as AHCI depend on that behavior to allocate system |
| 132 | structures for which the chipsets only allow one DMA descriptor. |
| 133 | -05:59- : alexh@: http://svn.freebsd.org/viewvc/base/head/sys/vm/redzone.c?view=markup&pathrev=155086 |
| 134 | < this is redzone. it basically calls redzone_addr_ntor() to increase the size in |
| 135 | malloc(), and then redzone_setup() just before returning the chunk |
| 136 | -06:02- : dillon@: jeeze. that looks horrible. |
| 137 | -06:03- : alexh@: I don't quite get that nsize + redzone_roundup(nsize) |
| 138 | -06:03- : dillon@: I don't get it either. It would completely break power-of-2-sized alignments in the |
| 139 | original request |
| 140 | -06:04- : dillon@: hmmm. well, no it won't break them, but the results are oging to be weird |
| 141 | -06:04- : dillon@: ick. |
| 142 | |
| 143 | -06:15- : dillon@: if the original request is a power of 2 the redzone adjusted request must be a power |
| 144 | of 2 |
| 145 | -06:15- : dillon@: basically |
| 146 | -06:16- : dillon@: so original request 64, redzone request must be 128, 256, 512, 1024, etc. |
| 147 | -06:16- : alexh@: yah, k |
| 148 | -06:16- : dillon@: original request 32, current redzone code would be 32+128 which is WRONG. |
| 149 | -06:16- : alexh@: how big is PAGE_SIZE ? |
| 150 | -06:16- : dillon@: 4096 on i386 and amd64 |
| 151 | -06:17- : alexh@: and one single malloc can't be bigger than that? |
| 152 | -06:17- : dillon@: I'm fairly sure our kmalloc does not guarantee alignment past PAGE_SIZE (that is, |
| 153 | the alignment will be only PAGE_SIZE eve if you allocate PAGE_SIZE*2) |
| 154 | -06:17- : dillon@: a single kmalloc can be larger then PAGE_SIZe |
| 155 | -06:18- : dillon@: it will use the zone up to around 1/2 the zone size (~64KB I think), after which it |
| 156 | allocates pages directly with the kernel kvm allocator |
| 157 | -06:18- : dillon@: if you look at the kmalloc code you will see the check for oversized allocations |
| 158 | -06:18- : alexh@: yah, saw that |
| 159 | -06:18- : alexh@: "handle large allocations directly" |
| 160 | -06:19- : alexh@: not sure how to do this, really, as the size is obviously also changed in |
| 161 | kmem_slab_alloc |
| 162 | -06:20- : alexh@: but kmem_slab_alloc isn't called always, is it? |
| 163 | -06:20- : alexh@: only if the req doesn't fit into an existant zone |
| 164 | -06:20- : dillon@: right |
| 165 | -06:20- : dillon@: you don't want to redzone the zone allocation itself |
| 166 | |
| 167 | </pre> |