Support for the TSO memory model on Arm CPUs
Support for the TSO memory model on Arm CPUs
Posted Apr 26, 2024 17:09 UTC (Fri) by flussence (guest, #85566)Parent article: Support for the TSO memory model on Arm CPUs
<s>
I understand the objection! There's plenty of other examples of bad code that deserves to fail on systems that do things correctly: turning memory overcommit on, for instance, encourages a whole lot of sloppy programming practices that aren't portable to other OSes. Or trying to play sound in the Flash plugin on glibc circa 2008 and expecting it to not come out garbled - who does that? Also nobody on Linux should expect their CPU scheduler to keep their cores 100% loaded, or their disk scheduler to let them do literally anything else with the system while writing to a USB stick; I'm sure they'll revert those patches real soon now and we can go back to pretending we're a Real UNIX™.
I understand the objection! There's plenty of other examples of bad code that deserves to fail on systems that do things correctly: turning memory overcommit on, for instance, encourages a whole lot of sloppy programming practices that aren't portable to other OSes. Or trying to play sound in the Flash plugin on glibc circa 2008 and expecting it to not come out garbled - who does that? Also nobody on Linux should expect their CPU scheduler to keep their cores 100% loaded, or their disk scheduler to let them do literally anything else with the system while writing to a USB stick; I'm sure they'll revert those patches real soon now and we can go back to pretending we're a Real UNIX™.
We must punish users of stinky software by making their lives miserable through sanctions - that will convince the upstream application developers, three steps removed on another architecture, OS and economic system, to fix it promptly so that we never have to leave our elitist comfort zone.
</s>
God forbid anyone gets functioning drivers for mainstream hardware in Linux.
