Messages in this thread | | | Subject | Re: Regression: hibernation is broken since e6bc9de714972cac34daa1dc1567ee48a47a9342 | From | Brad Campbell <> | Date | Tue, 17 Mar 2020 23:51:58 +0800 |
| |
On 26/2/20 4:26 am, Darrick J. Wong wrote: > On Sun, Feb 23, 2020 at 08:03:11PM +0100, Domenico Andreoli wrote: >> On Fri, Feb 21, 2020 at 04:23:19PM -0800, Darrick J. Wong wrote: >>> >>> Ok, third try. Does the following work? This is a little more >>> selective in that it only disables the write protection on the swap >>> device/file that uswusp is going to write to. >> >> Yes it works but also verified that once the S_SWAPFILE bit is cleared >> it's never restored, therefore the protecton is gone after the first >> hibernation. > > Ok, good. Now can you try the third part, which ought to re-apply > S_SWAPFILE after a successful resume, please? Assuming this works, I > think we're ready with a fixpatch. >
I just bumped up against it upgrading from 5.2 to 5.5 & a long bisection results in apparently the same point : # first bad commit: [dc617f29dbe5ef0c8ced65ce62c464af1daaab3d] vfs: don't allow writes to swap files
Tested-By: Brad Campbell <lists2009@fnarfbargle.com>
> --D > > diff --git a/include/linux/swap.h b/include/linux/swap.h > index 1e99f7ac1d7e..add93e205850 100644 > --- a/include/linux/swap.h > +++ b/include/linux/swap.h > @@ -458,6 +458,7 @@ extern void swap_free(swp_entry_t); > extern void swapcache_free_entries(swp_entry_t *entries, int n); > extern int free_swap_and_cache(swp_entry_t); > extern int swap_type_of(dev_t, sector_t, struct block_device **); > +extern void swap_relockall(void); > extern unsigned int count_swap_pages(int, int); > extern sector_t map_swap_page(struct page *, struct block_device **); > extern sector_t swapdev_block(int, pgoff_t); > diff --git a/kernel/power/user.c b/kernel/power/user.c > index 77438954cc2b..b11f7037ce5e 100644 > --- a/kernel/power/user.c > +++ b/kernel/power/user.c > @@ -271,6 +271,8 @@ static long snapshot_ioctl(struct file *filp, unsigned int cmd, > break; > } > error = hibernation_restore(data->platform_support); > + if (!error) > + swap_relockall(); > break; > > case SNAPSHOT_FREE: > @@ -372,10 +374,17 @@ static long snapshot_ioctl(struct file *filp, unsigned int cmd, > */ > swdev = new_decode_dev(swap_area.dev); > if (swdev) { > + struct block_device *bd; > + > offset = swap_area.offset; > - data->swap = swap_type_of(swdev, offset, NULL); > + data->swap = swap_type_of(swdev, offset, &bd); > if (data->swap < 0) > error = -ENODEV; > + > + inode_lock(bd->bd_inode); > + bd->bd_inode->i_flags &= ~S_SWAPFILE; > + inode_unlock(bd->bd_inode); > + bdput(bd); > } else { > data->swap = -1; > error = -EINVAL; > diff --git a/mm/swapfile.c b/mm/swapfile.c > index b2a2e45c9a36..a64dcba10db6 100644 > --- a/mm/swapfile.c > +++ b/mm/swapfile.c > @@ -1799,6 +1799,32 @@ int swap_type_of(dev_t device, sector_t offset, struct block_device **bdev_p) > return -ENODEV; > } > > +/* Re-lock swap devices after resuming from userspace suspend. */ > +void swap_relockall(void) > +{ > + int type; > + > + spin_lock(&swap_lock); > + for (type = 0; type < nr_swapfiles; type++) { > + struct swap_info_struct *sis = swap_info[type]; > + struct block_device *bdev = bdgrab(sis->bdev); > + > + /* > + * uswsusp only knows how to suspend to block devices, so we > + * can skip swap files. > + */ > + if (!(sis->flags & SWP_WRITEOK) || > + !(sis->flags & SWP_BLKDEV)) > + continue; > + > + inode_lock(bd->bd_inode); > + bd->bd_inode->i_flags |= S_SWAPFILE; > + inode_unlock(bd->bd_inode); > + bdput(bd); > + } > + spin_unlock(&swap_lock); > +} > + > /* > * Get the (PAGE_SIZE) block corresponding to given offset on the swapdev > * corresponding to given index in swap_info (swap type). >
| |