Hi Tom, On Mon, 14 Oct 2024 at 09:02, Tom Rini <[email protected]> wrote: > > On Mon, Oct 14, 2024 at 12:26:02PM +0530, Sughosh Ganu wrote: > > On Sun, 13 Oct 2024 at 23:21, Tom Rini <[email protected]> wrote: > > > > > > On Sun, Oct 13, 2024 at 04:25:09PM +0530, Sughosh Ganu wrote: > > > > > > > Add a flag LMB_NONOTIFY that can be passed to the LMB API's for > > > > reserving memory. This will then result in no notification being sent > > > > from the LMB module for the changes to the LMB's memory map. > > > > > > > > Signed-off-by: Sughosh Ganu <[email protected]> > > > > > > Can you please elaborate on what we need this flag for, for example what > > > the valid situation for it is? We should be updating (or adding comments > > > such that updates are automatic..) to doc/api/lmb.rst as well. Thanks. > > > > The flag is used to prevent notifying the EFI code about the change in > > the LMB memory map when the alloc/free request originates in the EFI > > memory module. > > Simon, this is the loop you referred to on IRC the other day, and > instead you solve this with the flag about a point of cooperation?
Yes, that's right. I'd like to have EFI allocations confined to the EFI area until efi_init_obj_list() is called, then have EFI add the lmb reservations to its own table, then own memory-allocation from then on. Basically if we apply my series[1] first, it will remove the need for this EFI call and the associated complexity. Most of this series will still be needed though. > > > The EFI memory module adds the memory region to its > > memory map as part of the alloc/free operation. The idea is that the > > LMB module should notify EFI about changes to its memory map when > > these changes are initiated from non-EFI consumers of memory. > > OK, so the first change is that since doc/api/lmb.rst is just a > kerneldoc of include/lmb.h we need to add the flag here, and we need to > first document LMB_NOOVERWRITE which is already undocumented. The next > change is that at the end of the series, update the rst with some > paragraphs about what and why, for clarity, after auditing include/lmb.h > for other items that are missing kerneldoc comments. I am willing to > make that second change a follow-up series, however. Regards, Simon [1] https://patchwork.ozlabs.org/project/uboot/list/?series=427729
