On Thu, Apr 9, 2020 at 3:45 AM Zhou Wang <wangzhou1@xxxxxxxxxxxxx> wrote: > > On 2020/4/9 0:27, Arnd Bergmann wrote: > > The added dependencies must be applied recursively to > > other modules that select CRYPTO_DEV_HISI_QM, to avoid > > running into the same problem again: > > > > WARNING: unmet direct dependencies detected for CRYPTO_DEV_HISI_QM > > Depends on [m]: CRYPTO [=y] && CRYPTO_HW [=y] && (ARM64 [=y] || COMPILE_TEST [=y]) && PCI [=y] && PCI_MSI [=y] && (UACCE [=m] || UACCE [=m]=n) > > Selected by [y]: > > - CRYPTO_DEV_HISI_SEC2 [=y] && CRYPTO [=y] && CRYPTO_HW [=y] && PCI [=y] && PCI_MSI [=y] && (ARM64 [=y] || COMPILE_TEST [=y] && 64BIT [=y]) > > - CRYPTO_DEV_HISI_HPRE [=y] && CRYPTO [=y] && CRYPTO_HW [=y] && PCI [=y] && PCI_MSI [=y] && (ARM64 [=y] || COMPILE_TEST [=y] && 64BIT [=y]) > > ld: drivers/crypto/hisilicon/qm.o: in function `hisi_qm_uninit': qm.c:(.text+0x23b8): undefined reference to `uacce_remove' > > > > Fixes: 47c16b449921 ("crypto: hisilicon - qm depends on UACCE") > > Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx> > > --- > > Seems we already have a fix: https://lkml.org/lkml/2020/3/30/173 with your reviewed-by :) Ah, of course. I even remembered that patch when I saw the problem in mainline, but for some reason assumed it was yet another instance of the same bug, rather than a patch that had I was just missing in my tree. Arnd