Re: [PATCH 2/3] libnvdimm: nd_region flush callback support

From: Pankaj Gupta
Date: Wed Sep 05 2018 - 04:40:17 EST

Next message: Jia-Ju Bai: "[PATCH v2] lightnvm: pblk-recovery: lightnvm: pblk: Fix two sleep-in-atomic-context bugs"
Previous message: Jia-Ju Bai: "Re: [PATCH] lightnvm: pblk: Fix two sleep-in-atomic-context bugs in pblk_line_submit_smeta_io()"
In reply to: kbuild test robot: "Re: [PATCH 2/3] libnvdimm: nd_region flush callback support"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Hello,

Thanks for the report.

>
> Hi Pankaj,
>
> Thank you for the patch! Perhaps something to improve:
>
> [auto build test WARNING on linux-nvdimm/libnvdimm-for-next]
> [also build test WARNING on v4.19-rc2 next-20180831]
> [if your patch is applied to the wrong git tree, please drop us a note to
> help improve the system]
>
> url:
> https://github.com/0day-ci/linux/commits/Pankaj-Gupta/kvm-fake-DAX-device/20180903-160032
> base: https://git.kernel.org/pub/scm/linux/kernel/git/nvdimm/nvdimm.git
> libnvdimm-for-next
> reproduce:
> # apt-get install sparse
> make ARCH=x86_64 allmodconfig
> make C=1 CF=-D__CHECK_ENDIAN__
> :::::: branch date: 7 hours ago
> :::::: commit date: 7 hours ago
>
> drivers/nvdimm/pmem.c:116:25: sparse: expression using sizeof(void)
> drivers/nvdimm/pmem.c:135:25: sparse: expression using sizeof(void)
> >> drivers/nvdimm/pmem.c:204:32: sparse: incorrect type in assignment
> >> (different base types) @@ expected restricted blk_status_t [usertype]
> >> bi_status @@ got e] bi_status @@

I will fix this in V2. Will wait for any review comments and address in v2.

Thanks,
Pankaj

> drivers/nvdimm/pmem.c:204:32: expected restricted blk_status_t
> [usertype] bi_status
> drivers/nvdimm/pmem.c:204:32: got int
> drivers/nvdimm/pmem.c:208:9: sparse: expression using sizeof(void)
> drivers/nvdimm/pmem.c:208:9: sparse: expression using sizeof(void)
> include/linux/bvec.h:82:37: sparse: expression using sizeof(void)
> include/linux/bvec.h:82:37: sparse: expression using sizeof(void)
> include/linux/bvec.h:83:32: sparse: expression using sizeof(void)
> include/linux/bvec.h:83:32: sparse: expression using sizeof(void)
> drivers/nvdimm/pmem.c:220:32: sparse: incorrect type in assignment
> (different base types) @@ expected restricted blk_status_t [usertype]
> bi_status @@ got e] bi_status @@
> drivers/nvdimm/pmem.c:220:32: expected restricted blk_status_t
> [usertype] bi_status
> drivers/nvdimm/pmem.c:220:32: got int
>
> #
> https://github.com/0day-ci/linux/commit/69b95edd2a1f4676361988fa36866b59427e2cfa
> git remote add linux-review https://github.com/0day-ci/linux
> git remote update linux-review
> git checkout 69b95edd2a1f4676361988fa36866b59427e2cfa
> vim +204 drivers/nvdimm/pmem.c
>
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 107
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 108 static
> void write_pmem(void *pmem_addr, struct page *page,
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 109 unsigned
> int off, unsigned int len)
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 110 {
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 111 unsigned
> int chunk;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 112 void
> *mem;
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 113
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 114 while
> (len) {
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 115 mem =
> kmap_atomic(page);
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 @116 chunk =
> min_t(unsigned int, len, PAGE_SIZE);
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 117
> memcpy_flushcache(pmem_addr, mem + off, chunk);
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 118
> kunmap_atomic(mem);
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 119 len -=
> chunk;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 120 off = 0;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 121 page++;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 122
> pmem_addr += PAGE_SIZE;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 123 }
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 124 }
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 125
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 126 static
> blk_status_t read_pmem(struct page *page, unsigned int off,
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 127 void
> *pmem_addr, unsigned int len)
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 128 {
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 129 unsigned
> int chunk;
> 60622d682 drivers/nvdimm/pmem.c Dan Williams 2018-05-03 130 unsigned
> long rem;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 131 void
> *mem;
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 132
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 133 while
> (len) {
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 134 mem =
> kmap_atomic(page);
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 135 chunk =
> min_t(unsigned int, len, PAGE_SIZE);
> 60622d682 drivers/nvdimm/pmem.c Dan Williams 2018-05-03 136 rem =
> memcpy_mcsafe(mem + off, pmem_addr, chunk);
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 137
> kunmap_atomic(mem);
> 60622d682 drivers/nvdimm/pmem.c Dan Williams 2018-05-03 138 if (rem)
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 139 return
> BLK_STS_IOERR;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 140 len -=
> chunk;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 141 off = 0;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 142 page++;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 143
> pmem_addr += PAGE_SIZE;
> 98cc093cb drivers/nvdimm/pmem.c Huang Ying 2017-09-06 144 }
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 145 return
> BLK_STS_OK;
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 146 }
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 147
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 148 static
> blk_status_t pmem_do_bvec(struct pmem_device *pmem, struct page *page,
> 3f289dcb4 drivers/nvdimm/pmem.c Tejun Heo 2018-07-18 149
> unsigned int len, unsigned int off, unsigned int op,
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 150
> sector_t sector)
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 151 {
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 152
> blk_status_t rc = BLK_STS_OK;
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 153 bool
> bad_pmem = false;
> 32ab0a3f5 drivers/nvdimm/pmem.c Dan Williams 2015-08-01 154
> phys_addr_t pmem_off = sector * 512 + pmem->data_offset;
> 7a9eb2066 drivers/nvdimm/pmem.c Dan Williams 2016-06-03 155 void
> *pmem_addr = pmem->virt_addr + pmem_off;
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 156
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 157 if
> (unlikely(is_bad_pmem(&pmem->bb, sector, len)))
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 158 bad_pmem
> = true;
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 159
> 3f289dcb4 drivers/nvdimm/pmem.c Tejun Heo 2018-07-18 160 if
> (!op_is_write(op)) {
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 161 if
> (unlikely(bad_pmem))
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 162 rc =
> BLK_STS_IOERR;
> b5ebc8ec6 drivers/nvdimm/pmem.c Dan Williams 2016-03-06 163 else {
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 164 rc =
> read_pmem(page, off, pmem_addr, len);
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 165
> flush_dcache_page(page);
> b5ebc8ec6 drivers/nvdimm/pmem.c Dan Williams 2016-03-06 166 }
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 167 } else {
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 168 /*
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 169 * Note
> that we write the data both before and after
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 170 *
> clearing poison. The write before clear poison
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 171 *
> handles situations where the latest written data is
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 172 *
> preserved and the clear poison operation simply marks
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 173 * the
> address range as valid without changing the data.
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 174 * In
> this case application software can assume that an
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 175 *
> interrupted write will either return the new good
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 176 * data
> or an error.
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 177 *
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 178 *
> However, if pmem_clear_poison() leaves the data in an
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 179 *
> indeterminate state we need to perform the write
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 180 * after
> clear poison.
> 0a370d261 drivers/nvdimm/pmem.c Dan Williams 2016-04-14 181 */
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 182
> flush_dcache_page(page);
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 183
> write_pmem(pmem_addr, page, off, len);
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 184 if
> (unlikely(bad_pmem)) {
> 3115bb02b drivers/nvdimm/pmem.c Toshi Kani 2016-10-13 185 rc =
> pmem_clear_poison(pmem, pmem_off, len);
> bd697a80c drivers/nvdimm/pmem.c Vishal Verma 2016-09-30 186
> write_pmem(pmem_addr, page, off, len);
> 59e647398 drivers/nvdimm/pmem.c Dan Williams 2016-03-08 187 }
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 188 }
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 189
> b5ebc8ec6 drivers/nvdimm/pmem.c Dan Williams 2016-03-06 190 return
> rc;
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 191 }
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 192
> dece16353 drivers/nvdimm/pmem.c Jens Axboe 2015-11-05 193 static
> blk_qc_t pmem_make_request(struct request_queue *q, struct bio *bio)
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 194 {
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 195
> blk_status_t rc = 0;
> f0dc089ce drivers/nvdimm/pmem.c Dan Williams 2015-05-16 196 bool
> do_acct;
> f0dc089ce drivers/nvdimm/pmem.c Dan Williams 2015-05-16 197 unsigned
> long start;
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 198 struct
> bio_vec bvec;
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 199 struct
> bvec_iter iter;
> bd842b8ca drivers/nvdimm/pmem.c Dan Williams 2016-03-18 200 struct
> pmem_device *pmem = q->queuedata;
> 7e267a8c7 drivers/nvdimm/pmem.c Dan Williams 2016-06-01 201 struct
> nd_region *nd_region = to_region(pmem);
> 7e267a8c7 drivers/nvdimm/pmem.c Dan Williams 2016-06-01 202
> d2d6364dc drivers/nvdimm/pmem.c Ross Zwisler 2018-06-06 203 if
> (bio->bi_opf & REQ_PREFLUSH)
> 69b95edd2 drivers/nvdimm/pmem.c Pankaj Gupta 2018-08-31 @204
> bio->bi_status = nd_region->flush(nd_region);
> 69b95edd2 drivers/nvdimm/pmem.c Pankaj Gupta 2018-08-31 205
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 206
> f0dc089ce drivers/nvdimm/pmem.c Dan Williams 2015-05-16 207 do_acct =
> nd_iostat_start(bio, &start);
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 208
> bio_for_each_segment(bvec, bio, iter) {
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 209 rc =
> pmem_do_bvec(pmem, bvec.bv_page, bvec.bv_len,
> 3f289dcb4 drivers/nvdimm/pmem.c Tejun Heo 2018-07-18 210
> bvec.bv_offset, bio_op(bio), iter.bi_sector);
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 211 if (rc)
> {
> 4e4cbee93 drivers/nvdimm/pmem.c Christoph Hellwig 2017-06-03 212
> bio->bi_status = rc;
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 213 break;
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 214 }
> e10624f8c drivers/nvdimm/pmem.c Dan Williams 2016-01-06 215 }
> f0dc089ce drivers/nvdimm/pmem.c Dan Williams 2015-05-16 216 if
> (do_acct)
> f0dc089ce drivers/nvdimm/pmem.c Dan Williams 2015-05-16 217
> nd_iostat_end(bio, start);
> 61031952f drivers/nvdimm/pmem.c Ross Zwisler 2015-06-25 218
> 1eff9d322 drivers/nvdimm/pmem.c Jens Axboe 2016-08-05 219 if
> (bio->bi_opf & REQ_FUA)
> 69b95edd2 drivers/nvdimm/pmem.c Pankaj Gupta 2018-08-31 220
> bio->bi_status = nd_region->flush(nd_region);
> 61031952f drivers/nvdimm/pmem.c Ross Zwisler 2015-06-25 221
> 4246a0b63 drivers/nvdimm/pmem.c Christoph Hellwig 2015-07-20 222
> bio_endio(bio);
> dece16353 drivers/nvdimm/pmem.c Jens Axboe 2015-11-05 223 return
> BLK_QC_T_NONE;
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 224 }
> 9e853f231 drivers/block/pmem.c Ross Zwisler 2015-04-01 225
>
> ---
> 0-DAY kernel test infrastructure Open Source Technology Center
> https://lists.01.org/pipermail/kbuild-all Intel Corporation
>

Next message: Jia-Ju Bai: "[PATCH v2] lightnvm: pblk-recovery: lightnvm: pblk: Fix two sleep-in-atomic-context bugs"
Previous message: Jia-Ju Bai: "Re: [PATCH] lightnvm: pblk: Fix two sleep-in-atomic-context bugs in pblk_line_submit_smeta_io()"
In reply to: kbuild test robot: "Re: [PATCH 2/3] libnvdimm: nd_region flush callback support"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]