So far the most absurd issue I've encountered on Linux / LS1028A/ DesignWare USB3 controller: the value of the kernel loglevel influences if USB works or not https://community.mnt.re/t/no-keyboard-input-with-ls1028-board-and-latest-image/1994/13
Conversation
Notices
-
minute (mntmn@mastodon.social)'s status on Tuesday, 04-Jun-2024 01:45:28 JST minute -
minute (mntmn@mastodon.social)'s status on Tuesday, 04-Jun-2024 02:31:07 JST minute @ruff yes, as I was writing. Or cache coherency issue.
In conversation permalink -
ruff@social.librem.one's status on Tuesday, 04-Jun-2024 02:31:08 JST ruff @mntmn that's indication of timing/race
In conversation permalink -
minute (mntmn@mastodon.social)'s status on Tuesday, 04-Jun-2024 05:43:28 JST minute @yannsionneau did dwc2 have a similar issue?
In conversation permalink -
Yann Sionneau (yannsionneau@mastodon.online)'s status on Tuesday, 04-Jun-2024 05:43:29 JST Yann Sionneau @mntmn oh, I see ... dwc3 is not any better than dwc2 then :')
Synopsis T_TIn conversation permalink -
minute (mntmn@mastodon.social)'s status on Tuesday, 04-Jun-2024 05:57:13 JST minute @dok @yannsionneau big, big oof
In conversation permalink -
Yann Sionneau (yannsionneau@mastodon.online)'s status on Tuesday, 04-Jun-2024 05:57:15 JST Yann Sionneau @mntmn not similar issues but issues ^^
Like for instance 8000 irq/sec whatever you do over USB (even : nothing)
In conversation permalink -
dok (dok@rivals.space)'s status on Tuesday, 04-Jun-2024 05:57:15 JST dok @yannsionneau @mntmn dwc2 cannot handle periodic transfers on its own, thus it fire an IRQ every Start of Frame (SOF) which is 8 times per ms (when running at High Speed).
This is only an issue when dwc2 is used with a host role, and if one of the connected device requires some interrupt or isochronous transfers.
In conversation permalink
-