ddf0676f1a
When instantiating a user-created PHB on P9/P10, we don't really have a reason any more to go through an indirection in pnv_chip_add_phb() in pnv.c, we can go straight to the right function in pnv_phb4_pec.c. That way, default PHBs and user-created PHBs are all handled in the same file. This patch also renames pnv_phb4_get_pec() to pnv_pec_add_phb() to better reflect that it "hooks" a PHB to a PEC. For P8, the PHBs are parented to the chip directly, so it makes sense to keep calling pnv_chip_add_phb() in pnv.c, to also be consistent with where default PHBs are handled. The only change here is that, since that function is now only used for P8, we can refine the return type. So overall, the PnvPHB front-end now has a pnv_phb_user_get_parent() function which handles the parenting of the user-created PHBs by calling the right function in the right file based on the processor version. It's also easily extensible if we ever need to support a different parent object. Signed-off-by: Frederic Barrat <fbarrat@linux.ibm.com> Message-Id: <20230302163715.129635-5-fbarrat@linux.ibm.com> Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com> Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com> |
||
---|---|---|
.. | ||
bonito.c | ||
designware.c | ||
dino.c | ||
gpex-acpi.c | ||
gpex.c | ||
grackle.c | ||
gt64120.c | ||
i440fx.c | ||
Kconfig | ||
meson.build | ||
mv643xx.h | ||
mv64361.c | ||
pam.c | ||
pnv_phb3_msi.c | ||
pnv_phb3_pbcq.c | ||
pnv_phb3.c | ||
pnv_phb4_pec.c | ||
pnv_phb4.c | ||
pnv_phb.c | ||
pnv_phb.h | ||
ppce500.c | ||
q35.c | ||
raven.c | ||
remote.c | ||
sabre.c | ||
sh_pci.c | ||
trace-events | ||
trace.h | ||
uninorth.c | ||
versatile.c | ||
xen_igd_pt.c | ||
xilinx-pcie.c |