1 ### Note this is my personal todo and ideas list, don't jump on me if you don't like something or think that it isn't important enough.
5 - update it part by part!
9 - port this: http://svn.freebsd.org/viewvc/base/head/sys/dev/kbd/kbd.c?revision=112050&view=markup
10 - remove kbd_reinit_struct
12 - remove need for mplock
15 * sync syscons as far as possible
16 - tty changes not possible...
20 - probably way too much effort.
22 * port the mpsafe tty stuff from FreeBSD - or mpsafe ours.
23 - http://svn.freebsd.org/viewvc/base?view=revision&revision=181905 and quite a few later ones
24 - http://p4db.freebsd.org/changeList.cgi?FSPC=%2F%2Fdepot%2Fprojects%2Fmpsafetty%2F...&ignore=GO%21
25 - all related drivers, listed in UPDATING in the above revision, and others
26 - DON'T port pts driver, rather stick to ours but modify it to work with new ttys
27 - http://svn.freebsd.org/viewvc/base/head/sys/dev/snp/snp.c?view=log
28 - http://svn.freebsd.org/viewvc/base/head/sys/dev/si/si.c?view=log
31 - wrapper is included for userland; should be easy to port
32 - http://svn.freebsd.org/viewvc/base?view=revision&revision=184610
35 o Added VT6105M specific register definitions. VT6105M has the
36 following hardware capabilities.
37 - Tx/Rx IP/TCP/UDP checksum offload.
38 - VLAN hardware tag insertion/extraction. Due to lack of information
39 for getting extracted VLAN tag in Rx path, VLAN hardware support
40 was not implemented yet.
41 - CAM(Content Addressable Memory) based 32 entry perfect multicast/
44 o Implemented CAM based 32 entry perfect multicast filtering for
45 VT6105M. If number of multicast entry is greater than 32, vr(4)
46 uses traditional hash based filtering.
48 * do something similar to geom (maybe with compat layer for geom?)
49 - http://svn.freebsd.org/viewvc/base?view=revision&revision=92108
50 - while here, "layerize" the disk subsystem
52 * continue work on disk scheduler
53 - focus on initial fair queueing policy as proof of concept
56 * ATA (automatic) spindown (see FreeBSD current)
58 * The ata(4) driver now supports a loader variable hw.ata.ata_dma_check_80pin.
59 - This can be used to disable the 80pin cable check on broken systems such as
60 certain laptops and Soekris boards. The default value is 1.
62 * RedZone, a buffer corruption protection for the kernel malloc(9) facility has been implemented.
63 - This detects both buffer underflows and overflows at runtime on free(9) and realloc(9),
64 and prints backtraces from where memory was allocated and from where it was freed.
67 * port uart driver (?)
69 * simplify/improve i386
70 - http://svn.freebsd.org/viewvc/base?view=revision&revision=173592
71 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/i686_mem.c?view=log
72 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/initcpu.c?view=log
73 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/pmap.c?view=log
74 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/locore.s?view=log
76 * suspend/resume for SMP x86
77 - http://lists.freebsd.org/pipermail/freebsd-acpi/2008-May/004879.html
81 * AMD64 suspend/resume
82 - http://svn.freebsd.org/viewvc/base?view=revision&revision=189903
84 * code simplification (Introduce cpu_vendor_id and replace a lot of strcmp(cpu_vendor, "...")
85 - http://svn.freebsd.org/viewvc/base?view=revision&revision=185341
88 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/mptable.c?view=log
89 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/mptable_pci.c?view=log
90 - http://svn.freebsd.org/viewvc/base/head/sys/i386/i386/io_apic.c?view=log
91 - MOST IMPORTANTLY: http://svn.freebsd.org/viewvc/base?view=revision&revision=121986
92 - AND: http://svn.freebsd.org/viewvc/base?view=revision&revision=121991
94 * port FreeBSD minidump, only kernel memory dump
95 - http://svn.freebsd.org/viewvc/base?view=revision&revision=176304
96 - http://svn.freebsd.org/viewvc/base?view=revision&revision=159121
97 - http://svn.freebsd.org/viewvc/base?view=revision&revision=157912
98 - http://svn.freebsd.org/viewvc/base?view=revision&revision=157911
99 - http://svn.freebsd.org/viewvc/base?view=revision&revision=157908
100 - http://svn.freebsd.org/viewvc/base?view=revision&revision=157909
102 * crash analysis tool
103 - http://svn.freebsd.org/viewvc/base?view=revision&revision=181335
106 - CFLAGS="-DUTMPX_USE_LIBRARY -DHAVE_UTMPX_H -DUSE_UTMPX"
107 - see branch, but still missing update to pam! (proper update)
110 http://svn.freebsd.org/viewvc/base?view=revision&revision=127969
113 - use namecache tricks as nullfs does
114 - make it work without whiteout
119 * synchronization routines documentation page (compare all the following, highlight differences)
124 - lwkt serialization tokens
132 devclass_find_internal(const char *classname, const char *parentname,
147 [alexh@leaf:~/home] $ roundup-server -p 8080 bt=bugtracker
150 -05:48- : dillon@: no, double frees to the object cache are nasty. It can't detect them. the object
151 winds up in the magazine array twice
152 -05:48- : dillon@: (and possibly different magazines, too)
153 -05:49- : alexh@: can't I just write some magic to a free object on the first objcache_put and check
154 if it's there on objcache_put?
155 -05:49- : alexh@: and clear it on objcache_get, anyways
156 -05:50- : dillon@: no, because the object is still may have live-initialized fields
157 -05:50- : dillon@: because it hasn't been dtor'ed yet (one of the features of the objcache, to avoid
158 having to reinitialize objects every time)
159 -05:50- : dillon@: the mbuf code uses that feature I think, probably other bits too
160 -05:51- : dillon@: theoretically we could allocate slightly larger objects and store a magic number at
161 offset [-1] or something like that, but it gets a little iffy doing that
162 -05:52- : dillon@: the objcache with the objcache malloc default could probably do something like that
164 -05:52- : dillon@: I don't consider memory tracking to be a huge issue w/ dragonfly, though I like the
165 idea of being able to do it. It is a much bigger problem in FreeBSD due to the
166 large number of committers
169 -05:55- : dillon@: For the slab allocator you may be able to do something using the Zone header.
170 -05:55- : dillon@: the slab allocator in fact I think already has optional code to allocate a tracking
171 bitmap to detect double-frees
172 -05:56- : dillon@: sorry, I just remembered the bit about the power-of-2 allocations
173 -05:56- : dillon@: for example, power-of-2-sized allocations are guaranteed not only to be aligned on
174 that particular size boundary, but also to not cross a PAGE_BOUNDARY (unless the
176 -05:57- : dillon@: various subsystems such as AHCI depend on that behavior to allocate system
177 structures for which the chipsets only allow one DMA descriptor.
178 -05:59- : alexh@: http://svn.freebsd.org/viewvc/base/head/sys/vm/redzone.c?view=markup&pathrev=155086
179 < this is redzone. it basically calls redzone_addr_ntor() to increase the size in
180 malloc(), and then redzone_setup() just before returning the chunk
181 -06:02- : dillon@: jeeze. that looks horrible.
182 -06:03- : alexh@: I don't quite get that nsize + redzone_roundup(nsize)
183 -06:03- : dillon@: I don't get it either. It would completely break power-of-2-sized alignments in the
185 -06:04- : dillon@: hmmm. well, no it won't break them, but the results are oging to be weird
186 -06:04- : dillon@: ick.
188 -06:15- : dillon@: if the original request is a power of 2 the redzone adjusted request must be a power
190 -06:15- : dillon@: basically
191 -06:16- : dillon@: so original request 64, redzone request must be 128, 256, 512, 1024, etc.
192 -06:16- : alexh@: yah, k
193 -06:16- : dillon@: original request 32, current redzone code would be 32+128 which is WRONG.
194 -06:16- : alexh@: how big is PAGE_SIZE ?
195 -06:16- : dillon@: 4096 on i386 and amd64
196 -06:17- : alexh@: and one single malloc can't be bigger than that?
197 -06:17- : dillon@: I'm fairly sure our kmalloc does not guarantee alignment past PAGE_SIZE (that is,
198 the alignment will be only PAGE_SIZE eve if you allocate PAGE_SIZE*2)
199 -06:17- : dillon@: a single kmalloc can be larger then PAGE_SIZe
200 -06:18- : dillon@: it will use the zone up to around 1/2 the zone size (~64KB I think), after which it
201 allocates pages directly with the kernel kvm allocator
202 -06:18- : dillon@: if you look at the kmalloc code you will see the check for oversized allocations
203 -06:18- : alexh@: yah, saw that
204 -06:18- : alexh@: "handle large allocations directly"
205 -06:19- : alexh@: not sure how to do this, really, as the size is obviously also changed in
207 -06:20- : alexh@: but kmem_slab_alloc isn't called always, is it?
208 -06:20- : alexh@: only if the req doesn't fit into an existant zone
209 -06:20- : dillon@: right
210 -06:20- : dillon@: you don't want to redzone the zone allocation itself