- i already have the unoptimized build made and just need to rebuild using it
- git_vector_get is a macro, but the data structure is not complex:
(gdb) p git_vector_get(>loose, iter->loose_pos)
No symbol "git_vector_get" in current context.
(gdb) p iter->loose
$14 = {_alloc_size = 3444, _cmp =
(gdb) n
877 refdb_fs_backend *backend =
GIT_CONTAINER_OF(iter->parent.db->backend, refdb_fs_backend, parent);
3: branch_name = "refs/tags/0.10.7"
(gdb) undisplay 3
(gdb) n
880 while (iter->loose_pos < iter->loose.length) {
(gdb) p (*iter)
$12 = {parent = {db =
Reposts supplied ' as is "
https://groups.io/g/marxmail/topic/the_big_bang_theory_is/92861144?p=,,,20,0,0,0::recentpostdate/sticky,,,20,2,0,92861144,previd%3D1659849313528521853,nextid%3D1659709020546104131=1659849313528521853=1659709020546104131
Unless you're a Marxist-Communist!
Karl Marx,
i added another display for branch_name, it's #3 now.
222 if ((error = iter->next(out, iter)) < 0)
3: branch_name = "refs/tags/0.10.7"
(gdb) s
refdb_fs_backend__iterator_next (out=0x555ac510, _iter=0x57fb0ec0)
at
reminder to self: the tag in quesiton is 0.10.7, and the goal is to
step into libgit, then step into iter->next .
the data structure is likely a linked list, so it would be good to
inspect it before it is advanced, so the two parts might be visible
side by side
Targeted, or political, assassinations are extrajudicial executions
https://groups.io/g/marxmail/topic/marjorie_cohn_biden_s/92865418?p=,,,20,0,0,0::recentpostdate/sticky,,,20,2,0,92865418,previd%3D1659849313528521853,nextid%3D1659709020546104131=1659849313528521853=1659709020546104131
Reposts
turns out the syntax is "undisplay 1".
i accidentally stepped over the next call, so i'm redoing it to get
back to the same state.
success. my undisplay call did not work.
(gdb) undisplay branch_name
(gdb) si
git_refdb_iterator_next (out=0x555ac510, iter=0x57fb0ec0)
at
/media/extradisk/src/codefudge/codefudge/datagen/bold84-cppgit2/ext/libgit2/src/refdb.c:219
219 {
1: branch_name = "refs/tags/0.10.7"
(gdb)
86 ++ reference_iter
1: branch_name = "refs/tags/0.10.7"
(gdb) s
repo_commits::reference_iterator::operator++ (this=0x555ac508) at
process2.cpp:253
253 if (git_reference_next(_ref, c_ptr) != 0) {
1: branch_name = "refs/tags/0.10.7"
(gdb) s
git_reference_next
The Reason Generator
"And in this room, we generate totally legitimate
[coughs]" I mean completely made up reasons for whatever Boss wants to happen."
The floating bundle of organs wearing a sign saying "tour guide"
gestures a part toward a pile of human beings with tubes and cables
coming out
here i am about to step into the iterator to find the next one:
(gdb) n
95 } catch (cppgit2::git_exception &) {
(gdb) n
89 auto branch_tip = branch.resolve().target();
(gdb) n
88 auto branch = *reference_iter;
(gdb) n
86 ++
(gdb) p branch
$6 = { = {}, c_ptr_ =
0x562f5d60, owner_ = cppgit2::ownership::libgit2}
(gdb) p *branch.c_ptr_
$7 = {db = 0x557b4aa0, type = GIT_REFERENCE_DIRECT, target = {oid
= {id = "\030O\035\237\262\351jy\207A¾\327\306?X\v\357$\020"},
symbolic = 0x796ae9b29f1d4f18 }, peel = {
(gdb) run
The program being debugged has been started already.
Start it from the beginning? (y or n) y
Starting program:
/media/extradisk/src/codefudge/codefudge/datagen/process2
../repos/Rust
[Thread debugging using libthread_db enabled]
Using host libthread_db library
current state of debugging:
I had encountered the branch name "refs/tags/0.10.7" twice in a row.
i'd like to step through the libgit structures and code moving from
the first one to the second, to gain information on how this happens
or what it means or indicates.
locally, i have a whole mess of public repositories squished together
into one repo, so as to work with their commits together, and i don't
yet know which one the commit came from. the function in question is
of course supposed to identify that.
this is the commit it happens to be looking at. theoretically, all the
named refs contain this commit. this hash is one way of finding the
repository containing the refs:
(gdb) p commit.to_hex_string(40)
$4 = "a3632327e441c5975f2e40ca794111c87de80d8a"
>>> Written by Karina Wiedman, who was born in Kazakhstan and has lived in
>>>Russia, Belarus and now the UK, The Anarchist is the inaugural winner of the
>>>Woven Voices prize for migrant writers <<<
so this is interesting: after refs/tags/0.10.7 it iterated for a bit
without hitting a breakpoint, and then encountered refs/tags/0.10.7
_again_
Breakpoint 2, repo_commits::remote_name[abi:cxx11](cppgit2::oid
const&) (this=0x555ac478, commit=...) at process2.cpp:96
96
it seems to be packing them fine now. it's taking a bit. 442.8MB
including the submodule dependencies.
i'll 'cont' a few more times and look at those tags.
We'll take a break when the perps neck breaks.
>>> Portland police announced on Thursday that they arrested 47-year-old
>>>Christopher Knipe on second-degree murder charges, for the 2019 murder of a
>>>23-year old anarchist, Sean “Armenio” Kealiher. <<<
the missing file is a symlink into the folder i deleted.
i'll review the build file, verify i am not linking with
tinytokenizers, and i suppose clean it or wipe its target directory or
whatnot
⠼ Packing 13339 files (393.9MB)Error: ENOENT: no such file or
directory, stat
'/media/extradisk/src/codefudge/codefudge/datagen/tinytokenizers/target/cxxbridge/tinytokenizers/src/tinytokenizers.rs.cc'
$ rm -rf tinytokenizers/target/release/build
$ w3 put .
i added breakpoints on lines 100 and 413, which are control paths via
which it can exit the iteration, so i can examine the state and
differentiate if it exits the loop correctly.
theoretically, i could hit 'cont' through every tag and see what happens.
meanwhile, i've been trying to pack the
i'm typing "cont" to keep hitting the breakpoints, and told it to
display the "branch_name" variable. it's enumerating tags that contain
the commit.
Continuing.
Breakpoint 1, repo_commits::remote_name[abi:cxx11](cppgit2::oid
const&) (this=0x555ac478, commit=...) at process2.cpp:92
92
for now i'm putting a breakpoint on line 96 too, inside the exception handler.
this lets me engage the situation in a way that doesn't freak me out,
seeing that it keeps returning to the debugger without any complaints,
as it goes over the references containing the commit
last time i used line 91 to break after the filter condition.
that's the static assignment and could have confused gdb, because it's
weird code that is run only on the first function call.
i'm breaking on line 92 instead.
anyway while that packs i'll resume troubleshooting the issue.
i know it's likely resolvable with some simple quirky bug that i just
need to find.
the code works with all offline data, so i could reduce variables, in
theory, by running offline
maybe i'll try to store it all
valgrind is not finding any errors yet, after running for some tens of seconds.
given last time i was hand-stepping, this likely indicates it won't
find any unless the loop terminates, changing the situation.
valgrind is still running .
at the launch of the script it has an unoptimized loop that indexes
every commit. there are a lot of them.
it just finished while typing this. now it is in the hang.
seems it would be simplest to enable ram dumping, and just cat the ram
and swap to an archive file. the ram dump could be mutated but it
would still be expected to explain the behavior in some way.
i don't see it in the list on wikipedia: although there is a key to
make the system crash, which then theoretically takes a crashdump,
maybe just of the kernel though. (alt- "sysrq", "c"; or somesuch)
i love imaging systems and storing the images! not what's up atm though.
thinking that
i think there might be a hotkey that makes a dump on linux, something
about sysrq near the concept of raising elephants
i mostly never work on security and engage systems so as to make them
_less_ secure due to my brainwashing
you can tell cause i don't use end to end email cryptography
for single systems, it would make sense to have a grub mode that
produces a hashed memory dump
you'd boot to the grub mode, photograph the hash, and upload the image
with the photo
this would not be bulletproof but would be exotic and strong enough to help
i am currently using this system to write this email
i do not have boot media prepared, to extract either the disk or the memory
i do not have a 2nd system on hand. there are also pci cards that can
extract memory; they likely also need a 2nd system.
this is a server i am running in the
note: one way to do that, is to suspend the system to disk, and then
boot off dedicated media and extract the memory from ram remanence.
this approach would fail if an attacker acted to deter it.
note: if this were caused by a trojan, malware, virus, etc, it would
be important to take a memory image with a 2nd system or dedicated
hardware
i'm websearching for "core dump of gdb" briefly
i get very confused around these things
> warning: Probes-based dynamic linker interface failed.
> Reverting to original interface.
I'm guessing the above quote is related to the following then
happening. The breakpoint was not hit, so I interrupted execution to
try to quickly find the outer code executing:
(gdb) cont
Continuing.
^C
so while valgrind finds that the error is my mind, not my system, i'll
paste some scrollback in
here's where i identified that the hang was within the remote_name()
function. i set a breakpoint within it and restarted execution:
410 auto remote_name =
On 8/7/22, Undiscussed Groomed for Male Slavery, One Victim of Many
wrote:
> ok, this time i stepped into the first call and it turns out the
> return statement is throwing an exception inside it, and gdb's "step"
> and "nexti" commands do not step into the exception handler, instead
>
notes: thread local name setting is line 92. handler is line 96.
handler appends to list.
i have things making sense now, having placed a breakpoint inside the
exception handler. finally gdb is behaving in a normal way around the
issue.
i guess it would make sense to run it under valgrind
ok, this time i stepped into the first call and it turns out the
return statement is throwing an exception inside it, and gdb's "step"
and "nexti" commands do not step into the exception handler, instead
continuing. the iterator theoretically steps to the next value and the
breakpoint is hit
process2.cpp is my code, so i'm imagining the most likely scenario
here is that i typed something into gdb that stimulated finding the
crash, and forgot when i walked away.
$ gdb --version
GNU gdb (Ubuntu 9.2-0ubuntu1~20.04.1) 9.2
$ b2sum $(type -p gdb)
e8a7e79ea79b2bc0977313899484136f34be865cd6032a26c1841a256d4b18a99b51dd8d0f05710ec5b3ee4da23041fc6872bd3650246bc0efc9fb050f902ff4
/usr/bin/gdb
$ ldd $(type -p gdb) | sed -ne 's/.* => \(.*\) (.*/\1/p' | xargs b2sum
i booted up a web browser so as to paste in gdb snippets, and went to
get some water while i waited as the system has heavy i/o slowing
things down as it collates git commits in the background
when i returned gdb had crashed, without any user interaction:
Breakpoint 6,
I'm debugging a slowdown in my data generator for the adapters for semibalanced
trees stuff.
I have a lightweight c++ iterator to loop through repository references and to
find which branch a random commit came from.
The hang is happening inside the function that uses this iterator.
I have a
>>> I am very hostile to this ideology. Being Danish, German Nazis invaded my
>>>country, killed our policemen in concentration camps and so on. My own
>>>paternal grandfather was forced to flee to Sweden due to his anti-Nazi
>>>activism. To be fair, he also has some Jewish ancestry which I
Capital, vol. 3, chapter 47: “ The possibility is here presented for definite
economic development taking place, depending, of course, upon favourable
circumstances, inborn racial characteristics, etc…”
Engels: (Marx and Engels, Selected Works, Vol. 3, p. 502.): “We regard economic
conditions
Tim May, lacking the cosmopolitan sensibilities of Assange, “the idea of trying
to be Julian Assange gives me the creeps.” “I’m not concerned about things like
that. Let the Africans kill each other,” May insists. “I don’t have those kinds
of political interests”
Thankfully Tim May is dead -
In order to manage the constant flow of clickbait and disinformation peddled
here by the likes of Jeff Berwick's bum-buddy, I'm calling for a massive effort
to save the West from a devastating first-strike launched against our strongest
encryption protecting trillions ( beat )
Full-spectrum
" . . . As reported by Science magazine, the scientists used a system comprised
of 512 GPUs to complete the same calculation developed by Google to demonstrate
it had passed the quantum supremacy milestone back in 2019. . . "
512 has long seemed to me to place a neat upper-bound on how many
" . . . It is, of course the right decision to request a FOIA to review the
impact of the NSA on NIST. NIST, have a responsibility to be open and
transparent about how they manage the process. . . .'
Fair enough - but lets not get carried away here.
Jul 20, 2015 - Julian Assange: To be honest,
" . . . a bad standard was identified early and kept out of products. NIST may
not have caught the problem . . . "
https://www.theverge.com/2013/9/11/4718694/how-far-did-the-nsa-go-to-weaken-cryptography-standards
NIST was criminally negligent, or just corruptly criminal, or both, in cahoots
" . . . NIST, have a responsibility to be open and transparent about how they
manage the process. . ." Yeah, right.
The NSA are so forthcoming I got reams of █ from them.
>>> Smid recalls. "NIST is part of the government and so is the NSA. The NSA
>>>has submitted candidate
-- Forwarded message -
From: Gunnar Larson
Date: Fri, Aug 5, 2022, 8:25 AM
Subject: xNY.io - Bank.org: New York State Saftey Concern
To:
Cc: cypherpunks , Harris, Adrienne A (DFS) <
adrienne.har...@dfs.ny.gov>, Weber, Richard (DFS) ,
Alexi Anania , , Reader, Shaun
Dear Madam
You say " quantum cryptographers are stealing a quarter of a billion Euros from
the European Commission " like that’s a bad thing.
https://blog.cr.yp.to/index.html
Reposts etc
" I advocate the overthrow of all government ( beat ) through peace and quiet "
Gary Snyder
So who wants to join the LAST revolution on earth, the one that’ll take down
ALL the governments?
Before Jim Bell went to prison, he suspected that most government officials
were corrupt. Three years
Marjorie Taylor Greene cheered on Alex Jones after Sandy Hook verdict
A forensic economist testified that Alex Jones began funneling $11,000 per day
into an alleged shell company around the time he was found liable by default in
defamation suits
>>> an exposition of the projection model in the most creative ideas about
space in contemporary mathematics such as twisters, quasicrystals, and quantum
topology. Robbin clarifies these esoteric concepts with understandable drawings
and diagrams.
Robbin proposes that the powerful role of
This was the latest-dated reference I found in the July archives with a
relevant subject line.
On July 22, 2022 3:09:36 AM EDT, "Undiscussed Horrific Abuse, One Victim of
Many" wrote:
>here's a recent stable code. this was before the 'fiction' approach i
>believe; classes could still be
I'm thinking of working a little on the algorithm for random access append only
data again, but I'm not sure where it is.
Scientists say they've debunked Google’s quantum supremacy claims once and for
all
https://share.newsbreak.com/1joxv01w
A team of scientists in China claim to have replicated the performance of
Google’s Sycamore quantum computer using traditional hardware, thereby
undermining the suggestion
67 matches
Mail list logo
