You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We should be careful about the results of nm, the associated file is quite enormous
(it can go to more than 1Mio, and mean size is around 500Kio for EACH compilation) EDIT : I optionalized this feature with the --checksize option to not overload the database
Right now, we are computing different sizes, but all rely on the same procedure:
https://github.com/TuxML/ProjetIrma/blob/dev/compilation/compiler.py#L256-L259
(the "raw" size of the file)
It's correct, but we miss an opportunity to gather much more information for free.
We should use
size
(UNIX command)https://elinux.org/Kernel_Size_Tuning_Guide#Measuring_the_kernel
in particular
'size vmlinux
We can also use
nm --size -r vmlinux
We can add two new columns
size_vmlinux
(it will be string)and
nm_size_vmlinux
(I suspect we have to store the information in a .log file)It should be noted that we only have to do additional measures for
vmlinux
(not the compressed ones).Some tests are needed (taking tinyconfig and some random configurations of 4.13.3 and 4.15)
The text was updated successfully, but these errors were encountered: