Re: [PATCH 6.1 000/167] 6.1.136-rc1 review

[Greg Kroah-Hartman]

On Wed, Apr 30, 2025 at 11:54:49AM -0400, Matthew Rosato wrote:
> 
> > 2)
> > Regressions on s390 with defconfig builds with gcc-13, gcc-8 and
> > clang-20 and clang-nightly toolchains on the stable-rc 6.1.136-rc1.
> > 
> > * s390, build
> >   - clang-20-defconfig
> >   - clang-nightly-defconfig
> >   - gcc-13-allmodconfig
> >   - gcc-13-defconfig
> >   - gcc-8-defconfig-fe40093d
> > 
> > Regression Analysis:
> >  - New regression? Yes
> >  - Reproducibility? Yes
> > 
> ...
> > drivers/s390/virtio/virtio_ccw.c:88:9: error: unknown type name 'dma64_t'
> >    88 |         dma64_t queue;
> >       |         ^~~~~~~
> > drivers/s390/virtio/virtio_ccw.c:95:9: error: unknown type name 'dma64_t'
> >    95 |         dma64_t desc;
> >       |         ^~~~~~~
> > drivers/s390/virtio/virtio_ccw.c:99:9: error: unknown type name 'dma64_t'
> >    99 |         dma64_t avail;
> >       |         ^~~~~~~
> > drivers/s390/virtio/virtio_ccw.c:100:9: error: unknown type name 'dma64_t'
> >   100 |         dma64_t used;
> >       |         ^~~~~~~
> > drivers/s390/virtio/virtio_ccw.c:109:9: error: unknown type name 'dma64_t'
> >   109 |         dma64_t summary_indicator;
> >       |         ^~~~~~~
> > drivers/s390/virtio/virtio_ccw.c:110:9: error: unknown type name 'dma64_t'
> >   110 |         dma64_t indicator;
> >       |         ^~~~~~~
> > drivers/s390/virtio/virtio_ccw.c: In function 'virtio_ccw_drop_indicator':
> > drivers/s390/virtio/virtio_ccw.c:370:25: error: implicit declaration
> > of function 'virt_to_dma64'; did you mean 'virt_to_page'?
> > [-Werror=implicit-function-declaration]
> >   370 |                         virt_to_dma64(get_summary_indicator(airq_info));
> >       |                         ^~~~~~~~~~~~~
> >       |                         virt_to_page
> > drivers/s390/virtio/virtio_ccw.c:374:28: error: implicit declaration
> > of function 'virt_to_dma32'; did you mean 'virt_to_page'?
> > [-Werror=implicit-function-declaration]
> >   374 |                 ccw->cda = virt_to_dma32(thinint_area);
> >       |                            ^~~~~~~~~~~~~
> >       |                            virt_to_page
> > drivers/s390/virtio/virtio_ccw.c: In function 'virtio_ccw_setup_vq':
> > drivers/s390/virtio/virtio_ccw.c:552:45: error: implicit declaration
> > of function 'u64_to_dma64' [-Werror=implicit-function-declaration]
> >   552 |                 info->info_block->l.queue = u64_to_dma64(queue);
> >       |                                             ^~~~~~~~~~~~
> > drivers/s390/virtio/virtio_ccw.c: In function 'virtio_ccw_find_vqs':
> > drivers/s390/virtio/virtio_ccw.c:654:9: error: unknown type name 'dma64_t'
> >   654 |         dma64_t *indicatorp = NULL;
> >       |         ^~~~~~~
> > cc1: some warnings being treated as errors
> 
> The virtio_ccw errors are caused by '[PATCH 6.1 033/167] s390/virtio_ccw: fix virtual vs physical address confusion'
> 
> Picking the following 2 dependencies would resolve the build error:
> 
> 1bcf7f48b7d4 s390/cio: use bitwise types to allow for type checking
> 8b19e145e82f s390/cio: introduce bitwise dma types and helper functions

I'm just going to drop all of these now and wait for a tested series to
be sent.

thanks,

greg k-h