On 9/5/25 9:43 AM, Jerome Brunet wrote:
Hello Jerome,
pci_epf_alloc_space() works like this:
If the user requests a BAR size that is smaller than the fixed-size BAR,
it will allocate space matching the fixed-size.
As in most cases, having a BAR larger than needed by an EPF driver is
still acceptable.
However, if the user requests a size larger than the fixed-size BAR,
as in your case, we will return an error, as we cannot fulfill the
user's request.
I don't see any alternative other than your/Damien's proposal above.
Unfortunately, all EPF drivers would probably need this same change.
It seems that pci-epf-ntb and pci-epf-vntb only use BAR0 (BAR_CONFIG) and
BAR0+BAR1 (BAR_CONFIG and BAR_DB) , so those should be OK on this
controller. NVMe EPF also seems to use only BAR0 and it specifically
handles fixed size BAR. It seems everything that is in the tree so far
managed to sidestep hitting fixed-size BAR4 problems on this hardware,
except for the test driver.
As it stands, a vNTB device needs 3 BARs minimum (CFG, DB and MW). The
NTB one may get away with with 2 BARs, with DB and MW sharing one.
I clearly missed the MW, thanks for pointing this out.
If you referring to Renesas about that BAR4, I did use it for vNTB.
It is indeed not upstream ... yet [1]
I think it is possible to have vNTB on 2 BARs with some tweaks, putting
CFG and DB on the same one.
[1]: https://lore.kernel.org/r/20250702-ntb-rcar-support-v3-2-4268d9c85eb7@xxxxxxxxxxxx
Nitpick, commit message, "Renesas R-Car Gen4" (Gen3 has a different PCIe
controller) . Please CC linux-renesas-soc@xxxxxxxxxxxxxxx on V3 , thank
you !
--
Best regards,
Marek Vasut
