On 10/28/15 22:29, Laszlo Ersek wrote: > This is version 2 of > <http://thread.gmane.org/gmane.comp.bios.edk2.devel/3424>. > > Public branch: > <https://github.com/lersek/edk2/commits/cpump_exitboot_v2>. > > As suggested by Jordan, I'll keep this at the front of my SMM-for-OVMF > series. I'm posting it now for getting reviews, but I won't commit it > just yet. > > No functional changes, just differently organized, as requested by > Jordan. See the notes on the individual patches. > > Cc: Chen Fan <[email protected]> > Cc: Jeff Fan <[email protected]> > Cc: Jordan Justen <[email protected]> > Cc: Michael Kinney <[email protected]> > > Thanks > Laszlo > > Laszlo Ersek (4): > UefiCpuPkg/CpuDxe: Avoid ExitBootServices() callback if there is no AP > UefiCpuPkg/CpuDxe: Prepare final HALT loop for APs > UefiCpuPkg/CpuDxe: Fix ExitBootServices() callback in the presence of > SMIs > UefiCpuPkg/CpuDxe: Release AP startup code after ExitBootServices() > > UefiCpuPkg/CpuDxe/CpuMp.h | 46 +++++- > UefiCpuPkg/CpuDxe/ApStartup.c | 150 ++++++++++++++++++-- > UefiCpuPkg/CpuDxe/CpuMp.c | 47 +++--- > 3 files changed, 210 insertions(+), 33 deletions(-) >
I'd like to remind prospective reviewers that Paolo's suggestion today (1) to simply accept that AP-raised SMIs take longer to handle (while we'd also decrease the timeout temporarily to mitigate the same) renders it possible to postpone this series. Namely, version 1 (and this current version, v2 as well) of this series were prompted when the *broadcast* SMI exposed a bug in the ExitBootServices() handler. See the issue marked as [1] in message (2), and the description of the test environment / patches in the same. Meaning, if we accept Paolo's idea (1) -- and I certainly welcome it! --, then the broadcast SMI falls away, and *then* this ExitBootServices() handler bug in UefiCpuPkg/CpuDxe can be left dormant for a while longer. An SMI triggered by the BSP will not wake the APs from their wait-for-SIPI state. (See the first paragraph of "PATCH v2 2/4".) I guess this is the path of least resistance, so I'm revoking this series for now. (1) http://thread.gmane.org/gmane.comp.bios.edk2.devel/3672/focus=3749 (2) http://thread.gmane.org/gmane.comp.bios.edk2.devel/3357 Thanks Laszlo _______________________________________________ edk2-devel mailing list [email protected] https://lists.01.org/mailman/listinfo/edk2-devel
