On Tue, 2020-09-08 at 20:09 +-1000, Stephen Rothwell wrote: +AD4- Hi all, +AD4- +AD4- After merging the akpm-current tree, today's linux-next build (x86+AF8-64 +AD4- allmodconfig) failed like this: +AD4- +AD4- drivers/xen/unpopulated-alloc.c: In function 'fill+AF8-list': +AD4- drivers/xen/unpopulated-alloc.c:30:9: error: 'struct dev+AF8-pagemap' has +AD4- no member named 'res'+ADs- did you mean 'ref'? +AD4- 30 +AHw- pgmap-+AD4-res.name +AD0- +ACI-Xen scratch+ACIAOw- +AD4- +AHw- +AF4AfgB+- +AD4- +AHw- ref +AD4- drivers/xen/unpopulated-alloc.c:31:9: error: 'struct dev+AF8-pagemap' has +AD4- no member named 'res'+ADs- did you mean 'ref'? +AD4- 31 +AHw- pgmap-+AD4-res.flags +AD0- IORESOURCE+AF8-MEM +AHw- IORESOURCE+AF8-BUSY+ADs- +AD4- +AHw- +AF4AfgB+- +AD4- +AHw- ref +AD4- drivers/xen/unpopulated-alloc.c:33:51: error: 'struct dev+AF8-pagemap' +AD4- has no member named 'res'+ADs- did you mean 'ref'? +AD4- 33 +AHw- ret +AD0- allocate+AF8-resource(+ACY-iomem+AF8-resource, +ACY-pgmap-+AD4-res, +AD4- +AHw- +AF4AfgB+- +AD4- +AHw- ref +AD4- In file included from include/asm-generic/memory+AF8-model.h:5, +AD4- from arch/x86/include/asm/page.h:76, +AD4- from arch/x86/include/asm/thread+AF8-info.h:12, +AD4- from include/linux/thread+AF8-info.h:38, +AD4- from arch/x86/include/asm/preempt.h:7, +AD4- from include/linux/preempt.h:78, +AD4- from include/linux/spinlock.h:51, +AD4- from include/linux/mmzone.h:8, +AD4- from include/linux/gfp.h:6, +AD4- from drivers/xen/unpopulated-alloc.c:3: +AD4- drivers/xen/unpopulated-alloc.c:53:35: error: 'struct dev+AF8-pagemap' +AD4- has no member named 'res'+ADs- did you mean 'ref'? +AD4- 53 +AHw- xen+AF8-pfn+AF8-t pfn +AD0- PFN+AF8-DOWN(pgmap-+AD4-res.start)+ADs- +AD4- +AHw- +AF4AfgB+- +AD4- include/linux/pfn.h:20:23: note: in definition of macro 'PFN+AF8-DOWN' +AD4- 20 +AHw- +ACM-define PFN+AF8-DOWN(x) ((x) +AD4APg- PAGE+AF8-SHIFT) +AD4- +AHw- +AF4- +AD4- drivers/xen/unpopulated-alloc.c:58:30: error: 'struct dev+AF8-pagemap' +AD4- has no member named 'res'+ADs- did you mean 'ref'? +AD4- 58 +AHw- release+AF8-resource(+ACY-pgmap-+AD4-res)+ADs- +AD4- +AHw- +AF4AfgB+- +AD4- +AHw- ref +AD4- drivers/xen/unpopulated-alloc.c:69:28: error: 'struct dev+AF8-pagemap' +AD4- has no member named 'res'+ADs- did you mean 'ref'? +AD4- 69 +AHw- release+AF8-resource(+ACY-pgmap-+AD4-res)+ADs- +AD4- +AHw- +AF4AfgB+- +AD4- +AHw- ref +AD4- fs/fuse/virtio+AF8-fs.c: In function 'virtio+AF8-fs+AF8-setup+AF8-dax': +AD4- fs/fuse/virtio+AF8-fs.c:838:9: error: 'struct dev+AF8-pagemap' has no member +AD4- named 'res'+ADs- did you mean 'ref'? +AD4- 838 +AHw- pgmap-+AD4-res +AD0- (struct resource)+AHs- +AD4- +AHw- +AF4AfgB+- +AD4- +AHw- ref +AD4- +AD4- Caused by commit +AD4- +AD4- b3e022c5a68c (+ACI-mm/memremap+AF8-pages: convert to 'struct range'+ACI-) +AD4- +AD4- interacting with commit +AD4- +AD4- 9e2369c06c8a (+ACI-xen: add helpers to allocate unpopulated memory+ACI-) +AD4- +AD4- from Linus' tree (in v5.9-rc4) and commit +AD4- +AD4- 7e833303db20 (+ACI-virtiofs: set up virtio+AF8-fs dax+AF8-device+ACI-) +AD4- +AD4- from the fuse tree. +AD4- +AD4- I have added the following patch which may require more work but at +AD4- least makes it all build. +AD4- +AD4- From: Stephen Rothwell +ADw-sfr+AEA-canb.auug.org.au+AD4- +AD4- Date: Tue, 8 Sep 2020 20:00:20 +-1000 +AD4- Subject: +AFs-PATCH+AF0- merge fix up for +ACI-mm/memremap+AF8-pages: convert to +AD4- 'struct +AD4- range'+ACI- +AD4- +AD4- Signed-off-by: Stephen Rothwell +ADw-sfr+AEA-canb.auug.org.au+AD4- +AD4- --- +AD4- drivers/xen/unpopulated-alloc.c +AHw- 15 +-+-+-+-+-+-+-+-+------- +AD4- fs/fuse/virtio+AF8-fs.c +AHw- 3 +--- +AD4- 2 files changed, 10 insertions(+-), 8 deletions(-) +AD4- +AD4- diff --git a/drivers/xen/unpopulated-alloc.c +AD4- b/drivers/xen/unpopulated-alloc.c +AD4- index 3b98dc921426..9fa7ce330628 100644 +AD4- --- a/drivers/xen/unpopulated-alloc.c +AD4- +-+-+- b/drivers/xen/unpopulated-alloc.c +AD4- +AEAAQA- -18,6 +-18,7 +AEAAQA- static unsigned int list+AF8-count+ADs- +AD4- static int fill+AF8-list(unsigned int nr+AF8-pages) +AD4- +AHs- +AD4- struct dev+AF8-pagemap +ACo-pgmap+ADs- +AD4- +- struct resource res+ADs- +AD4- void +ACo-vaddr+ADs- +AD4- unsigned int i, alloc+AF8-pages +AD0- round+AF8-up(nr+AF8-pages, +AD4- PAGES+AF8-PER+AF8-SECTION)+ADs- +AD4- int ret+ADs- +AD4- +AEAAQA- -27,10 +-28,10 +AEAAQA- static int fill+AF8-list(unsigned int nr+AF8-pages) +AD4- return -ENOMEM+ADs- +AD4- +AD4- pgmap-+AD4-type +AD0- MEMORY+AF8-DEVICE+AF8-GENERIC+ADs- +AD4- - pgmap-+AD4-res.name +AD0- +ACI-Xen scratch+ACIAOw- +AD4- - pgmap-+AD4-res.flags +AD0- IORESOURCE+AF8-MEM +AHw- IORESOURCE+AF8-BUSY+ADs- This is broken... it needs to be converted to 'struct range'. I'll take care of that when I respin the series. Sorry for the thrash it seems this is a new memremap+AF8-pages() user since the conversion patches landed.
