Suppose I have function fn somewhere within the .text section of an ELF64 executable. Is there a way to know at which offset (in bytes) from the start of the ELF file the fn function is located? Note that I don't need to know at which VA it was relocated at linking time, but its position within the ELF file.
Generally yes, if you can parse the ELF file directly or combine output from tools like objdump and readelf.
More specific: You can get the offset and virtual address of your .text section with 'readelf -S file' - write those down.
Further you can list symbols with 'readelf -s file', as long your executable is not stripped, and your function is visible (not static or in an anonymous namespace) then you should find your function and the virtual address of it.
Thus you can calculate the offset via
fn symbol offset = fn symbol VA - .text VA + .text offset
Thats assuming you want to do it "offline" with common tools. Its more difficult if you dont have access to the unstripped ELF file, and since only a part of the ELF File remains in memory, probably not possible without adding some information with "offline" tricks.
simply use objdump -F option
user#phoenix-amd64:~$ objdump -D -F /opt/phoenix/i486/heap-xxx -D | grep main
08048630 <__libc_start_main#plt> (File Offset: 0x630):
8048679: e8 b2 ff ff ff call 8048630 <__libc_start_main#plt> (File
Offset: 0x630)
080487d5 <main> (File Offset: 0x7d5):
The answer by Norbert Lange works for the functions that are listed in the symbol table of the ELF file. But static functions will not be present there, so even if e.g. GDB could find them (by using DWARF debug info), readelf -s won't.
In this case, you can use GDB. For example, let's find the offset of xfce_displays_helper_normalize_crtc in /usr/bin/xfsettingsd (that was my actual use case, thus this obscure choice of an example).
$ gdb -q -ex 'p &xfce_displays_helper_normalize_crtc' -ex q xfsettingsd
Reading symbols from xfsettingsd...
Reading symbols from /usr/lib/debug/.build-id/b2/2ad9713642253d4d7a6f94acf0174ccfe3d487.debug...
$1 = (void (*)(XfceRRCrtc *, XfceDisplaysHelper *)) 0x11e80 <xfce_displays_helper_normalize_crtc>
Note that here we only load the file with GDB, don't let it start. And then use p command (print in full form) to get the address. So in my case, the function is at offset 0x11e80.
In some cases GDB will resolve the offset to virtual address even before we start or starti the program. This happens, in particular, on x86-32. In this case we can simply subtract the virtual address of the file image, given by readelf -l:
$ readelf -l /bin/sleep | grep ' VirtAddr \|\<LOAD *0x[0-9a-f]\+\>'
Type Offset VirtAddr PhysAddr FileSiz MemSiz Flg Align
LOAD 0x000000 0x08048000 0x08048000 0x05230 0x05230 R E 0x1000
In the example above, the virtual address of the file image is 0x8048000, which would have to be subtracted from virtual address of the function if GDB happens to output it instead of the offset.
Related
I am trying to run an octave function from a batch file. The function is well written given how it works when launched from within the Octave GUI.
The batch file, other than pointing to the octave function, defines the only argument needed by it.
A while back this was not a function but a simple Octave script and the commands used were ok.
The only issue I am encountering now is being able to pass the variable calculated by the batch file onto the octave function.
I have recently written an octave function to do some file management. It requires only one input from the user:
function replace_TMM (file_base)
where file_base is a string to specify what directory I am working on. So it has to be something like "Z:" or "I:" or so on.
I am quite sure that the function is well written since I am able to use it from Octave GUI without any issues.
The fact is that I would like to run this function from a batch file. Inside this batch file I wrote:
SET a=%cd:~0,2%
This command is able to identify the working directory so "a" will be equal to "Z:" or similars.
Now my issue is telling the batch file to evaluate the octave function using "a" as its input argument.
I tried stuff like:
"C:\Program Files\GNU Octave\Octave-7.3.0\mingw64\bin\octave-cli.exe" -q --eval _03_REPLACE_V04("'%cd:~0,2%'")
which does not seem to work. This kind of solution gives a syntax error at batch level, it is not even able to enter the octave file.
If I instead try something like:
"C:\Program Files\GNU Octave\Octave-7.3.0\mingw64\bin\octave-cli.exe" -q _03_REPLACE_V04.m Z:
It is able to enter the octave file but it does not process the function, just skips over it to get to the end of the script.
Same goes if I try the following:
"C:\Program Files\GNU Octave\Octave-7.3.0\mingw64\bin\octave-cli.exe" _03_REPLACE_V04.m -"Z:"
In brief I bvelieve that the function itself works, it is only a matter of passing a variable from the batch to the octave.
Would really appreciate some help, thanks in advance.
UPDATE 1
I have done what was suggested by #Dariush Gavari and used the following syntax:
"C:\Program Files\GNU Octave\Octave-7.3.0\mingw64\bin\octave-cli.exe" -q --eval "replace_TMM('%a%')"
This gets me the following error message:
error: 'replace_TMM' undefined near line 1, column 1
I believed that it was because it was not ablòe to find the script containing the function. This is saved in a file called _03_REPLACE_V04.m
For this reason I have tried with
"C:\Program Files\GNU Octave\Octave-7.3.0\mingw64\bin\octave-cli.exe" -q --eval _03_REPLACE_V04.m "replace_TMM('%a%')"
Leading to the following error:
error: --eval "CODE" and script file are mutually exclusive options
usage: octave [-HVWdfhiqvx] [--debug] [--doc-cache-file file] [--echo-commands]
[--eval CODE] [--exec-path path] [--experimental-terminal-widget]
[--gui] [--help] [--image-path path] [--info-file file]
[--info-program prog] [--interactive] [--line-editing] [--no-gui]
[--no-history] [--no-init-file] [--no-init-path] [--no-line-editing]
[--no-site-file] [--no-window-system] [--norc] [-p path]
[--path path] [--persist] [--server] [--silent] [--traditional]
[--verbose] [--version] [file]
I believed that the problem could also have been having the functional nd the file with two different names. To solve this I have kept the same file name but changed the function to match it:
function _03_REPLACE_V04 (file_base)
Then in the batch:
"C:\Program Files\GNU Octave\Octave-7.3.0\mingw64\bin\octave-cli.exe" -q --eval "_03_REPLACE_V04('%a%')"
Leading to:
warning: function '_03_REPLACE_V04' defined within script file '\99_TOOLS\OCTAVE_FILES\_03_REPLACE_V04.m'
error: invalid call to script \99_TOOLS\OCTAVE_FILES\_03_REPLACE_V04.m
error: called from
_03_REPLACE_V04
In other words still no way of making it work. :)
Octave provides the argv function, which returns a cellstring array of all arguments passed to the octave executable at the time of launch, or in the case where it was used to launch a script, then this is the script's arguments.
So presumably all you have to do to get your directory from within octave is argv(){1}
If you would like to convert your filename to an absolute filename, you could also do this from within octave via the make_absolute_filename function.
Incidentally, a very useful command you should know of in octave is the lookfor command. Writing lookfor arguments in the terminal returns a list with all functions which have the word "arguments" in their description; argv is at the top of that list.
You can then do help argv to see more details on that command.
I have downloaded the "NYU Depth V2" dataset and toolbox from here. In the toolbox there is a script called get_synched_frames.m. I do not have Matlab, so I have tried running it in Octave. Unfortunately, it does not work as expected.
The line
% Faster than matlab's Dir function for big directories and slow
% distributed file systems...
files = regexp(ls(sceneDir), '(\s+|\n)', 'split');
gives only
files =
{
[1,1] = a-1300302776.479149-3987628315.dump
}
but ls(sceneDir) shows all files in the directory. Has anyone experienced this?
The difference is probably not in regexp, but in the return value of ls. ls does not behave the same way in Matlab and Octave when you capture its return value. Matlab's ls returns a char row vector (single string as char) with multiple files listed in it as a multi-line string with embedded newlines; Octave's ls returns a 2-D char array with one file per line. (IMHO Octave's format is better; it is very difficult to parse Matlab's ls output in a reliably correct manner. (That regexp code is not adequate.))
You might just want this in Octave:
files = cellstr(ls(sceneDir));
I have the output of recursive grep (actually ag) in a buffer, which is of the form filename:linenumber: ... [match] ..., and I want to be able to go to the occurrence (file and line number) currently under the cursor. This told me that I could execute normal-mode movements, so after extracting the file:line portion, I wrote this function:
function OpenFileNewTab(name)
let l:pair=split(a:name, ":")
execute "tabnew" get(l:pair, 0)
execute "normal!" get(l:pair, 1) . "G"
endfunction
It is supposed to open the specified file in a tab and then do <lineno>G, like I am able to do manually, to go to the specified line number. However, the cursor just stays on line 1. What am I doing wrong?
This question, by title alone, would be an exact duplicate, but it talks locating symbols in other files, while I already have the locations at hand.
Edit: My mappings for grep / ag are as follows:
nnoremap <Leader>ag :execute "new \| read !ag --literal -w" "<C-r><C-w>" g:repo \| :set filetype=c<CR>
nnoremap <Leader>gf ^v2t:"zy :execute OpenFileNewTab("<C-r>z")<CR>
To get my grep / ag results, I put the cursor on the word I want to search and enter <leader>ag, then, in the new buffer, I put the cursor on a line and enter <leader>gf - it selects from the start up to the second colon and calls OpenFileNewTab.
Edit 2: I'm on Cygwin, if it is of any importance - I doubt it.
Why don't you set &grepprg to call ag ?
" according to man ag
set grepprg=ag\ --vimgrep\ $*
set grepformat=%f:%l:%c:%m
" And then (not tested)
nnoremap <Leader>ag :grep -w <c-r><c-w><cr>
As others have said in the comments, you are just trying to emulate what the quickfix windows already provides. And, we are lucky vim can call grep, and it has a variation point to let us specify which grep program we wish to use: 'grepprg'.
Use file-line plugin. Pressing Enter on a line in the quicklist will normally open that file; file-line will make any filename of the form file:line:column (and several other formats) to open file and position to line and column.
I only found this (old) thread after I posted the exact same question on vi.stackexchange: https://vi.stackexchange.com/q/39557/44764. To help anyone who comes looking, I post the best answer to my question below as an alternative to the answers already given.
The gF command, like gf, opens the file in a new tab but additionally it also positions the cursor on the line after the colon. (I note the OP defines <leader>gf so maybe vim/neovim didn't auto-define gf or gF at the time this thread was originally created.)
I am trying to convert some tcl script into a C++ program. I don't have much experience with tcl and am hoping someone could explain what some of the following things are actually doing in the tcl script:
1) set rtn [true_test_sfm $run_dir]
2) cd [glob $run_dir]
3) set pwd [pwd]
Is the first one just checking if true_test_sfm directory exists in run_dir?
Also, I am programming on a windows machine. Would the system function be the equivalent to exec statements in tcl? And if so how would I print the result of the system function call to stdout?
In Tcl, square brackets indicate "evaluate the code between the square brackets". The result of that evaluation is substituted for the entire square-bracketed expression. So, the first line invokes the function true_test_sfm with a single argument $run_dir; the result of that function call is then assigned to the variable rtn. Unfortunately, true_test_sfm is not a built-in Tcl function, which means it's user-defined, which means there's no way we can tell you what the effect of that function call will be based on the information you've provided here.
glob is a built-in Tcl function which takes a file pattern as an argument and then lists files that match that pattern. For example, if a directory contains files "foo", "bar" and "baz", glob b* would return a list of two files, "bar" and "baz". Therefore the second line is looking for any files that match the pattern given by $run_dir, then using the cd command (another Tcl built-in) to change to the directory found by glob. Probably $run_dir is not actually a file pattern, but an explicit file name (ie, no globbing characters like * or ? in the string), otherwise this code may break unexpectedly. On Windows, some combination of FindFirstFile/FindNextFile in C++ could be used as a substitute for glob in Tcl, and SetCurrentDirectory could substitute for cd.
pwd is another built-in Tcl function which returns the process current working directory as an absolute path. So the last line is querying the current working directory and saving the result in a variable named pwd. Here you could use GetCurrentDirectory as a substitute for pwd.
So the utility Diff works just like I want for 2 files, but I have a project that requires comparisons with more than 2 files at a time, maybe up to 10 at a time. This requires having all those files side by side to each other as well. My research has not really turned up anything, vimdiff seems to be the best so far with the ability to compare 4 at a time.
My question: Is there any utility to compare more than 2 files at a time, or a way to hack diff/vimdiff so it can do multiple comparisons? The files I will be comparing are relatively short so it should not be too slow.
Displaying 10 files side-by-side and highlighting differences can be easily done with Diffuse. Simply specify all files on the command line like this:
diffuse 1.txt 2.txt 3.txt 4.txt 5.txt 6.txt 7.txt 8.txt 9.txt 10.txt
Vim can already do this:
vim -d file1 file2 file3
But you're normally limited to 4 files. You can change that by modifying a single line in Vim's source, however. The constant DB_COUNT defines the maximum number of diffed files, and it's defined towards the top of diff.c in versions 6.x and earlier, or about two thirds of the way down structs.h in versions 7.0 and up.
diff has built-in option --from-file and --to-file, which compares one operand to all others.
--from-file=FILE1
Compare FILE1 to all operands. FILE1 can be a directory.
--to-file=FILE2
Compare all operands to FILE2. FILE2 can be a directory.
Note: argument name --to-file is optional.
e.g.
# this will compare foo with bar, then foo with baz .html files
$ diff --from-file foo.html bar.html baz.html
# this will compare src/base-main.js with all .js files in git repo,
# that has 'main' in their filename or path
$ git ls-files :/*main*.js | xargs diff -u --from-file src/base-main.js
Checkout "Beyond Compare": http://www.scootersoftware.com/
It lets you compare entire directories of files, and it looks like it runs on Linux too.
if your running multiple diff's based off one file you could probably try writing a script that has a for loop to run through each directory and run the diff. Although it wouldn't be side by side you could at least compare them quickly. hope that helped.
Not answering the main question, but here's something similar to what Benjamin Neil has suggested but diffing all files:
Store the filenames in an array, then loop over the combinations of size two and diff (or do whatever you want).
files=($(ls -d /path/of/files/some-prefix.*)) # Array of files to compare
max=${#files[#]} # Take the length of that array
for ((idxA=0; idxA<max; idxA++)); do # iterate idxA from 0 to length
for ((idxB=idxA + 1; idxB<max; idxB++)); do # iterate idxB + 1 from idxA to length
echo "A: ${files[$idxA]}; B: ${files[$idxB]}" # Do whatever you're here for.
done
done
Derived from #charles-duffy's answer: https://stackoverflow.com/a/46719215/1160428
There is a simple an good way to do this = GREP.
Depending on the size of the text you can copy and paste it, or you can redirect the input of the file to the grep command. If you make a grep -vir /path to make a reverse search or a grep -ir /path. This is my way for certification exams.