Ah, I see. <mutters something about non- sortable commit IDs making "earlier" difficult to ascertain; sighs in futility>
Thanks, Ravi (rpokala@) -----Original Message----- From: John Baldwin <j...@freebsd.org <mailto:j...@freebsd.org>> Date: Monday, March 10, 2025 at 11:15 To: Ravi Pokala <rpok...@freebsd.org <mailto:rpok...@freebsd.org>>, <src-committ...@freebsd.org <mailto:src-committ...@freebsd.org>>, <dev-commits-src-...@freebsd.org <mailto:dev-commits-src-...@freebsd.org>>, <dev-commits-src-main@FreeBSD.org <mailto:dev-commits-src-main@FreeBSD.org>> Subject: Re: git: 78cd83e4017b - main - devclass_alloc_unit: Go back to using M_WAITOK On 3/10/25 13:41, Ravi Pokala wrote: > Hi John, > >> This reverts commit 234683726708cf5212d672d676d30056d4133859. > > The commit message for that change says that this allocation might be done > while holding a lock, and so M_WAITOK is not valid. So then why is this > change okay? > > Thanks, Because the earlier commit to sdhci in this same push fixed it to create devices in a sleepable context. -- John Baldwin
