tree: https://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git
for-next
head: 3d7b481b8a996e68339268487c0d3c8e54a2c274
commit: 32c8d6969d43c1bce394f75e9bcf32697146a773 [16/17] gfs2: Don't get stuck
on partial buffered writes
config: hexagon-randconfig-r022-20220325
(https://download.01.org/0day-ci/archive/20220326/202203260944.dmj6xxb2-...@intel.com/config)
compiler: clang version 15.0.0 (https://github.com/llvm/llvm-project
0f6d9501cf49ce02937099350d08f20c4af86f3d)
reproduce (this is a W=1 build):
wget
https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O
~/bin/make.cross
chmod +x ~/bin/make.cross
#
https://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git/commit/?id=32c8d6969d43c1bce394f75e9bcf32697146a773
git remote add gfs2
https://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2.git
git fetch --no-tags gfs2 for-next
git checkout 32c8d6969d43c1bce394f75e9bcf32697146a773
# save the config file to linux build tree
mkdir build_dir
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross W=1
O=build_dir ARCH=hexagon SHELL=/bin/bash fs/gfs2/
If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <l...@intel.com>
All warnings (new ones prefixed by >>):
>> fs/gfs2/file.c:1073:14: warning: comparison of distinct pointer types
>> ('typeof ((1UL << 18) - ((unsigned long)(iocb->ki_pos) & ~(~((1 << 18) -
>> 1)))) *' (aka 'unsigned long *') and 'typeof (iov_iter_count(from)) *' (aka
>> 'unsigned int *')) [-Wcompare-distinct-pointer-types]
min_size = min(PAGE_SIZE - offset_in_page(iocb->ki_pos),
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
include/linux/minmax.h:45:19: note: expanded from macro 'min'
#define min(x, y) __careful_cmp(x, y, <)
^~~~~~~~~~~~~~~~~~~~~~
include/linux/minmax.h:36:24: note: expanded from macro '__careful_cmp'
__builtin_choose_expr(__safe_cmp(x, y), \
^~~~~~~~~~~~~~~~
include/linux/minmax.h:26:4: note: expanded from macro '__safe_cmp'
(__typecheck(x, y) && __no_side_effects(x, y))
^~~~~~~~~~~~~~~~~
include/linux/minmax.h:20:28: note: expanded from macro '__typecheck'
(!!(sizeof((typeof(x) *)1 == (typeof(y) *)1)))
~~~~~~~~~~~~~~ ^ ~~~~~~~~~~~~~~
1 warning generated.
vim +1073 fs/gfs2/file.c
1005
1006 static ssize_t gfs2_file_buffered_write(struct kiocb *iocb,
1007 struct iov_iter *from,
1008 struct gfs2_holder *gh)
1009 {
1010 struct file *file = iocb->ki_filp;
1011 struct inode *inode = file_inode(file);
1012 struct gfs2_inode *ip = GFS2_I(inode);
1013 struct gfs2_sbd *sdp = GFS2_SB(inode);
1014 struct gfs2_holder *statfs_gh = NULL;
1015 size_t prev_count = 0, window_size = 0;
1016 size_t read = 0;
1017 ssize_t ret;
1018
1019 /*
1020 * In this function, we disable page faults when we're holding
the
1021 * inode glock while doing I/O. If a page fault occurs, we
indicate
1022 * that the inode glock may be dropped, fault in the pages
manually,
1023 * and retry.
1024 */
1025
1026 if (inode == sdp->sd_rindex) {
1027 statfs_gh = kmalloc(sizeof(*statfs_gh), GFP_NOFS);
1028 if (!statfs_gh)
1029 return -ENOMEM;
1030 }
1031
1032 gfs2_holder_init(ip->i_gl, LM_ST_EXCLUSIVE, 0, gh);
1033 retry:
1034 ret = gfs2_glock_nq(gh);
1035 if (ret)
1036 goto out_uninit;
1037 retry_under_glock:
1038 if (inode == sdp->sd_rindex) {
1039 struct gfs2_inode *m_ip = GFS2_I(sdp->sd_statfs_inode);
1040
1041 ret = gfs2_glock_nq_init(m_ip->i_gl, LM_ST_EXCLUSIVE,
1042 GL_NOCACHE, statfs_gh);
1043 if (ret)
1044 goto out_unlock;
1045 }
1046
1047 current->backing_dev_info = inode_to_bdi(inode);
1048 pagefault_disable();
1049 ret = iomap_file_buffered_write(iocb, from, &gfs2_iomap_ops);
1050 pagefault_enable();
1051 current->backing_dev_info = NULL;
1052 if (ret > 0) {
1053 iocb->ki_pos += ret;
1054 read += ret;
1055 }
1056
1057 if (inode == sdp->sd_rindex)
1058 gfs2_glock_dq_uninit(statfs_gh);
1059
1060 if (should_fault_in_pages(ret, from, &prev_count,
&window_size)) {
1061 size_t min_size, leftover;
1062
1063 /*
1064 * Make sure to fault in enough memory to fill at least
one
1065 * page cache page. Otherwise, we could end up with a
partial
1066 * write that __iomap_write_end() treats as a
zero-length
1067 * write, and we would get stuck.
1068 *
1069 * Note that we assume that after
fault_in_iov_iter_readable(),
1070 * at least min_size bytes of memory will be readable.
This
1071 * could change with sub-page pointer color probing.
1072 */
> 1073 min_size = min(PAGE_SIZE - offset_in_page(iocb->ki_pos),
1074 iov_iter_count(from));
1075 if (window_size < min_size)
1076 window_size = min_size;
1077
1078 gfs2_holder_allow_demote(gh);
1079 leftover = fault_in_iov_iter_readable(from,
window_size);
1080 gfs2_holder_disallow_demote(gh);
1081 if (window_size - leftover >= min_size) {
1082 if (gfs2_holder_queued(gh))
1083 goto retry_under_glock;
1084 if (read && !(iocb->ki_flags & IOCB_DIRECT))
1085 goto out_uninit;
1086 goto retry;
1087 }
1088 }
1089 out_unlock:
1090 if (gfs2_holder_queued(gh))
1091 gfs2_glock_dq(gh);
1092 out_uninit:
1093 gfs2_holder_uninit(gh);
1094 if (statfs_gh)
1095 kfree(statfs_gh);
1096 return read ? read : ret;
1097 }
1098
--
0-DAY CI Kernel Test Service
https://01.org/lkp
