initThumbprints() now takes an application tag argument
so x509 and ssh can coexist.
the thumbprint entries can now hold both sha1 and sha256
hashes. okThumbprint() now takes a len argument for the
hash length used.
the new function okCertificate() hashes the certificate
with both and checks for any matches.
on failure, okCertificate() returns 0 and sets error string.
we also check for include loops now in thumbfiles, limiting
the number of includes to 8.
previously nusb/ptp showed only 'generic folder' association type as directories, but all associations work as directories. tested with LG G5/android 7.0
when we initiate re-key exchange we data packets can still
come in. so instead we have everything that can come in all
the time in dispatch() function (including KEXINIT) and have
the receiver process just call that in a loop. exclude dispatch
and the sender proc from corrupting each others sendpkt() calls
with the QLock sl.
the event library doesnt do any flow control. if the host
keeps sending data while the user is mousing arround,
extract() will collect the data until memory fills up.
so instead we abandon the event library and convert
the program to use channels.
this makes implementing ssh-rsa authentication easier, as we
then can convert the public key directly to ssh format and check
if the server will accept that public key.
tlshand just needs the modulus to see if the public key matches
the one it has from the certificate.
we used to check for mb->d->name != nil before stating, to avoid
stating pop3/imap mailboxes who don't really have a local file,
but this breaks when the md->d is reconstructed (faked) from
the index! resulting in the mailbox stop being refreshed.
the solution is to not have mb->d == nil for imap/pop mailoxes.
- ignore directories in dirskip()
- use sortkey of 0 for invalid items, otherwise it could confuse qsort()
- use file size from dirreadall() instead of doing stat
- various cleanups
theres a race condition when mail delivery to mdir is slow,
then upas/fs sees partial mail file and caches the truncated
file size.
to avoid this, delivery will create the new mail file with
the .tmp extension (which is ignored by upas/fs) and after
everything has been written, rename it to the final name.
calculate alloc flag before waserror(), as compilers like
gcc will not notice the value changing later because
setjump() restores the old value due to callee-saves.
change is applies here to make it easier to merge with
drawterm.
thanks to aiju for debugging this; used to cause drawterm
memory leak until compiled with gcc -O0.
when converting mpint to bytes, always pad it to the size of
the modulus (RSA,DHE,ECDHE). mptobytes() now takes a byte len
parameter which the caller usually calculates from the group
modulus using mpsignif(). this bug sometimes caused "bad record mac"
after the handshake.
use a shared buffer, given that msgSend()/msgRecv() don't overlap
we can use the first half for sending, and the top half for
receiving, shifting down as neccesary. the space beween sendp and
recvp is free.
explicitely check for overflow in msgSend().
usbcmd() with Rh2d used to return the command size (8+ndata) wile returning
only ndata for Rd2h. this changes it to always return ndata for Rh2d. it
mostly doesnt matter as Rh2d callers only check r < 0 for error, but this
makes the interface symmetic.
turns out on real hardware, the front falls off if we write
the completion queue doorbell registers without consuming
an entry. so only write the register when we have processed
something.
timerdel() did not make sure that the timer function
is not active (on another cpu). just acquiering the
Timer lock in the timer function only blocks the caller
of timerdel()/timeradd() but not the other way arround
(on a multiprocessor).
this changes the timer code to track activity of
the timer function, having timerdel() wait until
the timer has finished executing.
basic NVMe controller driver, reads and writes work.
"namespaces" show up as logical units.
uses pin/msi interrupts (no msi-x support yet).
one submission queue per cpu, shared completion queue.
no recovery from fatal controller errors.
only tested in qemu (no hardware available).
commiting this so it can be found by someone who has
hardware.
Once the Barnes-Hut tree is constructed, the gravitational
force calculations can be done in parallel by dividing the
bodies up between a number of procs.
on thinkpad x1v4, the PCMP structure resides in upper reserved memory
pa=0xd7f49000 - while system memory ends at 0x0ffff000; so we have to
vmap() it instead of KADDR().
the RSD structure for ACPI might reside in low memory, so we sould
KADDR() in that case.
- smtpcram() was replying with the challenge instead of the response... m(
- simplify doauth, use error string for auth_getuserpasswd() error logging
- enable #pragma varargck for dBprint()
The Plan 9 assemblers use strtoll to parse the integer literals
in their input. It turns out that this is almost correct, but
VLONG_MIN is clamped. This patch changes to use strtoull
in order to allow the full range of integers.
The plan 9 assemblers support the DUPOK flag on text symbols. They parse and
ignore it on GLOBL symbols. This patch makes it work in the linkers.
The reason I ran into this is because my programming language (Myrddin) uses
data symbols to generate type information, and it's useful to avoid
duplicating all of the type info in every file that gets generated.
devmouse controls the screen blanking timeout, so move the
code there avoiding cross calls between modules. the only
function that needs to be provided is blankscreen(), which
gets called with drawlock locked.
the blank timeout is set thru /dev/mousectl now, so kernels
without devvga can set it.
blanking now only happens while /dev/mouse is read. so this
avoids accidentally blanking the screen on cpu servers that
do not have a mouse to unblank it.
now that locking works, its time to get rid of it.
given that the only concurrency is between the periodic reader
process and the filesystem code, just make sure the filesystem
rpc handlers do not run concurrenty while syncing is in progress
with a single qlock.
Zooming when far away from the center of gravity
of the galaxy would zoom the center of the screen
out of view. Now adjust the origin so that the
center of the screen stays centered
- bind /n/src9 only when theres .hg directory, so you can use /root as /n/src9
- skip *.6 and *.orig files in source direcrtories
- skip *.pdf and *.ps files in /sys/man
- skip cpurc.local and termrc.local
- dont create /fd, /mnt and/n in the root
- be specific about /lib
- create /cfg/pxe
aiju → i don't like it, it's more bullshit ways to expose the key :)
aiju → if someone can grab /adm/keyseed, they can also grab /adm/users and /adm/keys
the assert fails for regcompnl(".") as TANY is compiled to
one instruction instead of two when nl == 0.
its not a problem when we end up with less instructions, so
changing the assert condition from == to <= to make sure we
didnt overrun the buffer.
--
cinap
the kernel will automatically timeout the eap blocked state
for us and reassociate for us, so there is no need to provoke
the ap to start the wpa/eap handshake.
make the polling faster changing the sleep time from 500 to
100 ms.
- add some milisecond timestamps to the status change debug printing
- flush the packets in the queue on deassoc to avoid processing old pae
packets on next association.
- make roaming timeout shorter (60 -> 20 seconds)
- automatically timeout and restart wpa/pae blocked state
- fix printing race when essid gets changed underneath seprint
from openbsd driver, it seems the Centrino Advanced-N 6030 and 6235
cards share the same device revision as the 6205 (Type6005). Also
changing the device revision field from 4 to 5 bits.
- drivers enable short preamble and sort timeslot depending
on the ap beacon capinfo field (bss->cap)
- wifi sets short preamble bit in capinfo on association request
- wifi sets short timeslot bit when ap advertized it in beacon
reverting asn1mpint() as all users really just expect
unsigned integers here. also openssl seems to interpret
rsa modulus as unsigned no matter what... so keeping
it as it was before.
handle nil cipher bytes in factotum_rsa_decrypt() due
to pkcs1padbuf() failing.
apply some lessions from intels berzerk paper:
instead of parsing the decrypted digest info blob, we
generate the *expected* blob's for all digest algorithms
that match the digest size and compare the results.
provide pkcs1 pad and unpad functions that consistently
enforce minimum padding size and handles block types 1
and 2.
quick fix is to bias the rounding so the msb will always
be zero. should write proper conversion code to actually
deal with signed mpints... also for asn1mpint()... -- cinap
Make the logic around who has priority over the final
match simpler by merging the priority generation and
match fields in a smarter way. Move the creation of
new thread matches up to the top to avoid jumping all
over the place.
don't deadlock when cwgrow() detects we'r out of worm space
by releasing the cache superblock buffer.
don't allocate space beyond the worm device when dumping by
having slit() check. after cwrecur(), we check if there are
enougth blocks remaining to write the dump date directories
and superblocks.
- no need to splhi() in timerset, always called with
interrupts off.
- make timerset always update the period (next == 0)
- remove period update in fastticks(), simplify
delta calculation.
given that we only pass uchar* with constant offsets
to the s and d arguments of ENCRYPT(), we do not need
the temporary variables sp/dp and the compiler is
smart enougth to combine the const offset with the ones
from GET4() and PUT4() and emit single load and store
instructions for the byte accesses.
introducing new ctrunc() function that invalidates any caches
for the passed in chan, invoked when handling wstat with a
specified file length or on file creation/truncation.
test program to reproduce the problem:
#include <u.h>
#include <libc.h>
#include <libsec.h>
void
main(int argc, char *argv[])
{
int fd;
Dir *d, nd;
fd = create("xxx", ORDWR, 0666);
write(fd, "1234", 4);
d = dirstat("xxx");
assert(d->length == 4);
nulldir(&nd);
nd.length = 0;
dirwstat("xxx", &nd);
d = dirstat("xxx");
assert(d->length == 0);
fd = open("xxx", OREAD);
assert(read(fd, (void*)&d, 4) == 0);
}
allocating AX,CX,DX last improves 64-bit multiplication-add
chains like a*b + c*d as the multiplication does not need to save
and restore AX and DX registers in most cases. reserving CX for
shifts also helps.
encode printable characters litteraly, use D() macro to
reduce DATA instruction overhead.
this halves the time spend assembling $CONF.root.s for
the kernel build.
the root fileserver is mounted with the mount cache enabled
causing the contents of the cfsctl file to be cached as well
leading to wrong results. so after we generated the new stat
buffer, we increment the qid.vers so the cache starts out
clean.
given that the igfx driver doesnt provide any acceleration functions
and drawing is usually faster with double buffering as it eleminates
reads over the pci bus, enable softscreen by default.
don't use rand() and scale it to 0..1, instead call
native frand() which produces uniform random number.
instead of seeding the rng with time(0), use truerand().
on 386 kernel, each processor has its own pdb where the primary
pdb for kernel mappings is on cpu0 and other cpu's lazily pull
pdb entries from cpu0 when they fault in vmapsync().
so we have to edit the table tables in the pdb of cpu0 and not
the current processor.
on some modern machines like the x250, the bios arranges the mtrr's
and the framebuffer membar in a way that doesnt allow us to mark
the framebuffer pages as write combining, leading to slow graphics.
since the pentium III, the processor interprets the page table bit
combinations of the WT, CD and bit7 bits as an index into the
page attribute table (PAT).
to not change the semantics of the WT and CD bits, we preserve
the bit patterns 0-3 and use the last entry 7 for write combining.
(done in mmuinit() for each core).
the new patwc() function takes virtual address range and changes
the page table marking the range as write combining. no attempt
is made on invalidating tlb's. doesnt matter in our case as the
following mtrr() call in screen.c does it for us.
the assumption of only one producer ((abs)moustratrack()) is not true
for external mouse events from /dev/mousein, so protect the mouse state
and queue with ilock().
get rid of mousecreate(), just use devcreate().
reset cursor when all instances of /dev/mouse and /dev/cursor got closed,
instead of also considering /dev/mousectl. the reason is that kbdfs keeps
the mousectl file open. so exiting a program that has the cursor changed
will properly reset the cursor to arrow.
don't access user buffer while holding cursor spinlock! the memory access
can fault. theres also no lock needed there, we'r just copying *from* the
cursor memory.
fix use of strtol(), p will always be set, check for end of string.
keep pointer coordinates onscreen (off by one).
make lastms() function to get the last millisecond delta of last
call for resynchronization.
fix msg[3] buffer overflow in m5mouseputc().
get rid of mouseshifted logic, it is not used.
theres a bug is in sclose() where it doesnt check if wp is beyond
the buffer. also wp was not updated after realloc().
bug was reported by porlock on 9fans:
Plan 9's implementation of the standard C functions snprintf and
vsnprintf have a buffer overrun bug.
If the buffer length equals the output length (without the terminating
null), then one too many characters is written to the buffer.
For example,
snprintf(buf, 4, "ABCD");
will write 5 characters to buf.
