Skip to content

very very minor comments. have to still go through the linker session #1

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: globals
Choose a base branch
from
Open

very very minor comments. have to still go through the linker session #1

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
19 changes: 15 additions & 4 deletions _posts/2016-11-11-globals.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,11 +6,10 @@ author: Nir Friedman and Sandeep Kumar Gupta

Doing a global correctly in C++ is not a trivial thing, when the global itself
is non-trivial. Or, put another way: globals that have constructors and
destructors can easily make a big mess. There's surprisingly few resources
destructors can easily make a big mess. There are surprisingly few resources
summarizing exactly what considerations are relevant in order to make sure
globals behave themselves. There are some well known cases, like static
initialization order fiasco (SIOF). Maybe this is because to really talk about
globals, we have to talk about the linker, and the linker is a strange beast.
globals behave themselves. There are some well known cases, like [static initialization order fiasco][siof]
. Maybe this is because to really talk about globals, we have to talk about the linker, and the linker is a strange beast.
Without it, you can't write any real C++ program, but it's not covered by the
standard, its behavior is relatively rarely discussed, and varies from platform
to platform.
Expand Down Expand Up @@ -72,6 +71,9 @@ There are three translation units (TUs) here, one for each file ending in
`.cpp`. As far as a compiler is concerned, all of these translation units are
kosher. Let's try to link this together and see what happens:

==> Not sure how this looks in final page, but they look pretty bad in markdown page in browser. How about making
==> this as they appear in terminal?

{% highlight bash linenos %}
# Compile static translation unit
clang++ -std=c++14 -c -fpic static.cpp
Expand Down Expand Up @@ -275,3 +277,12 @@ Key point: assembly analysis shows that with static linked first, executable
initializes global, otherwise it doesn't. Shared library will initialize it
either way. Because the symbol of the global itself is already defined before
either initialization occurs, the initialization occurs on the same symbol.

[[siof]]: https://isocpp.org/wiki/faq/ctors#static-init-order


[Comment to be removed]
* I found this article http://www.lurklurk.org/linkers/linkers.html while searching for whether c++ standard talks about linking or not. Not sure how useful it is.
* Do you think it makes sense to run main.x.cpp under address sanitizer to see how it performs?
* The other thing that I noticed is usage of both "we need to understand" and "you need to understand". The latter seems pedantic, so prefer first?
* To see all the symbols in an executable I find this command handy: `readelf -s main.x.o | c++filt`. It lists undefined symbols as UND