1 ## Note: this is my personal todo and ideas list (alexh@)
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
7 * bring the samba3 hammer shadow copy foo to maturity
9 * add a communication channel mechanism to dm
10 - essentially a way to send messages and receive responses to dm target instances
12 * add another optimized bcopy/memcpy/memset/bzero version, using the AVX instructions
14 * add serpent and twofish xts modes to opencrypto
16 * add serpent and twofish support to dm_target_crypt
18 * Add TrueCrypt support
19 - Simply add a userland tool that sets dm_target_crypt up with the right parameters, extracted from the TrueCrypt header
20 - https://github.com/bwalex/tc-play <- done!
21 - need passphrase read timeout support
23 - cryptdisks: needs generic crypto options; multiple keyfile support
27 * Keep opencrypto up-to-date
28 - http://www.openbsd.org/cgi-bin/cvsweb/src/sys/crypto/
30 * Port hwpmc & dig into (boot-up) performance
34 - separate out common arch parts (linprocfs, for example)
36 * Fix the crash analysis script (or rather the programs it calls [some segfault])
38 * Take a look at updating lvm/dm/libdevicemapper
40 * Take a look at importing libdm from NetBSD
41 - would allow tc-play and similarly newly developed foo to avoid GPL license issues
43 * rip out the disk partitioning from the disk subsystem and implement it in a more general fashion
44 - crazy idea: as dm targets with an auto-configuration option!
45 - would require to be able to create dm targets with an arbitrary name and not in /dev/mapper
47 * ATA (automatic) spindown (see FreeBSD current)
50 http://svn.freebsd.org/viewvc/base?view=revision&revision=127969
54 - some incorrect accounting going on, don't remember details :)
57 - make it work without whiteout
63 * Fix ipsec, get rid of old ipsec
66 - Added VT6105M specific register definitions. VT6105M has the following hardware capabilities.
67 - Tx/Rx IP/TCP/UDP checksum offload.
68 - VLAN hardware tag insertion/extraction. Due to lack of information
69 for getting extracted VLAN tag in Rx path, VLAN hardware support
70 was not implemented yet.
71 - CAM(Content Addressable Memory) based 32 entry perfect multicast/
74 o Implemented CAM based 32 entry perfect multicast filtering for
75 VT6105M. If number of multicast entry is greater than 32, vr(4)
76 uses traditional hash based filtering.
78 * RedZone, a buffer corruption protection for the kernel malloc(9) facility has been implemented.
79 - This detects both buffer underflows and overflows at runtime on free(9) and realloc(9),
80 and prints backtraces from where memory was allocated and from where it was freed.
83 * port uart driver (?)
85 * port wscons (?) or update syscons
86 - probably way too much effort (wscons)
90 - wrapper is included for userland; should be easy to port
91 - http://svn.freebsd.org/viewvc/base?view=revision&revision=184610
92 - http://turbocat.net/~hselasky/usb4bsd/
93 - http://gitweb.dragonflybsd.org/~polachok/dragonfly.git/shortlog/refs/heads/usb2
95 * suspend/resume for SMP x86
96 - http://lists.freebsd.org/pipermail/freebsd-acpi/2008-May/004879.html
98 * AMD64 suspend/resume
99 - http://svn.freebsd.org/viewvc/base?view=revision&revision=189903
106 [alexh@leaf:~/home] $ roundup-server -p 8080 bt=bugtracker
109 -05:48- : dillon@: no, double frees to the object cache are nasty. It can't detect them. the object
110 winds up in the magazine array twice
111 -05:48- : dillon@: (and possibly different magazines, too)
112 -05:49- : alexh@: can't I just write some magic to a free object on the first objcache_put and check
113 if it's there on objcache_put?
114 -05:49- : alexh@: and clear it on objcache_get, anyways
115 -05:50- : dillon@: no, because the object is still may have live-initialized fields
116 -05:50- : dillon@: because it hasn't been dtor'ed yet (one of the features of the objcache, to avoid
117 having to reinitialize objects every time)
118 -05:50- : dillon@: the mbuf code uses that feature I think, probably other bits too
119 -05:51- : dillon@: theoretically we could allocate slightly larger objects and store a magic number at
120 offset [-1] or something like that, but it gets a little iffy doing that
121 -05:52- : dillon@: the objcache with the objcache malloc default could probably do something like that
123 -05:52- : dillon@: I don't consider memory tracking to be a huge issue w/ dragonfly, though I like the
124 idea of being able to do it. It is a much bigger problem in FreeBSD due to the
125 large number of committers
128 -05:55- : dillon@: For the slab allocator you may be able to do something using the Zone header.
129 -05:55- : dillon@: the slab allocator in fact I think already has optional code to allocate a tracking
130 bitmap to detect double-frees
131 -05:56- : dillon@: sorry, I just remembered the bit about the power-of-2 allocations
132 -05:56- : dillon@: for example, power-of-2-sized allocations are guaranteed not only to be aligned on
133 that particular size boundary, but also to not cross a PAGE_BOUNDARY (unless the
135 -05:57- : dillon@: various subsystems such as AHCI depend on that behavior to allocate system
136 structures for which the chipsets only allow one DMA descriptor.
137 -05:59- : alexh@: http://svn.freebsd.org/viewvc/base/head/sys/vm/redzone.c?view=markup&pathrev=155086
138 < this is redzone. it basically calls redzone_addr_ntor() to increase the size in
139 malloc(), and then redzone_setup() just before returning the chunk
140 -06:02- : dillon@: jeeze. that looks horrible.
141 -06:03- : alexh@: I don't quite get that nsize + redzone_roundup(nsize)
142 -06:03- : dillon@: I don't get it either. It would completely break power-of-2-sized alignments in the
144 -06:04- : dillon@: hmmm. well, no it won't break them, but the results are oging to be weird
145 -06:04- : dillon@: ick.
147 -06:15- : dillon@: if the original request is a power of 2 the redzone adjusted request must be a power
149 -06:15- : dillon@: basically
150 -06:16- : dillon@: so original request 64, redzone request must be 128, 256, 512, 1024, etc.
151 -06:16- : alexh@: yah, k
152 -06:16- : dillon@: original request 32, current redzone code would be 32+128 which is WRONG.
153 -06:16- : alexh@: how big is PAGE_SIZE ?
154 -06:16- : dillon@: 4096 on i386 and amd64
155 -06:17- : alexh@: and one single malloc can't be bigger than that?
156 -06:17- : dillon@: I'm fairly sure our kmalloc does not guarantee alignment past PAGE_SIZE (that is,
157 the alignment will be only PAGE_SIZE eve if you allocate PAGE_SIZE*2)
158 -06:17- : dillon@: a single kmalloc can be larger then PAGE_SIZe
159 -06:18- : dillon@: it will use the zone up to around 1/2 the zone size (~64KB I think), after which it
160 allocates pages directly with the kernel kvm allocator
161 -06:18- : dillon@: if you look at the kmalloc code you will see the check for oversized allocations
162 -06:18- : alexh@: yah, saw that
163 -06:18- : alexh@: "handle large allocations directly"
164 -06:19- : alexh@: not sure how to do this, really, as the size is obviously also changed in
166 -06:20- : alexh@: but kmem_slab_alloc isn't called always, is it?
167 -06:20- : alexh@: only if the req doesn't fit into an existant zone
168 -06:20- : dillon@: right
169 -06:20- : dillon@: you don't want to redzone the zone allocation itself