Is It Time To Overhaul The GNU Dynamic Linker?
At the GNU Tools Track during this week's Linux Plumbers Conference was an interesting talk by Red Hat's Ben Woodard. He shares his perspective on how the GNU dynamic linker (ld.so) could be ripe for an overhaul in the 2020's.
Given the stature today of Linux and the Unix wars being over, Woodard expresses his views on it could be time to overhaul the dynamic linker/loader and make it much more fitted for today's worlds. Possible areas for improvement include more robust linking, faster linking, and new performance and debugging angles. Among his "crazy ideas" would be experimenting with an ABI-aware loader, tool interface registration with the loader, an early fork in the loader for introducing new behavior, a modular library requirement solver, and more.
No code is written, at least not publicly yet, for an overhauled GNU dynamic linker/loader but given Red Hat's engineering resources it will be interesting to see if this is an area they pursue moving forward.
Those interested in the topic can see this PDF slide deck.
Given the stature today of Linux and the Unix wars being over, Woodard expresses his views on it could be time to overhaul the dynamic linker/loader and make it much more fitted for today's worlds. Possible areas for improvement include more robust linking, faster linking, and new performance and debugging angles. Among his "crazy ideas" would be experimenting with an ABI-aware loader, tool interface registration with the loader, an early fork in the loader for introducing new behavior, a modular library requirement solver, and more.
No code is written, at least not publicly yet, for an overhauled GNU dynamic linker/loader but given Red Hat's engineering resources it will be interesting to see if this is an area they pursue moving forward.
Those interested in the topic can see this PDF slide deck.
40 Comments