ore/exofs: Define new ore_verify_layout

All users of the ore will need to check if current code
supports the given layout. For example RAID5/6 is not
currently supported.

So move all the checks from exofs/super.c to a new
ore_verify_layout() to be used by ore users.

Note that any new layout should be passed through the
ore_verify_layout() because the ore engine will prepare
and verify some internal members of ore_layout, and
assumes it's called.

Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>

diff --git a/fs/exofs/inode.c b/fs/exofs/inode.c
index 96366a1..5a62420 100644 (file)
--- a/fs/exofs/inode.c
+++ b/fs/exofs/inode.c
@@ -37,11 +37,7 @@
 
 #define EXOFS_DBGMSG2(M...) do {} while (0)
 
-enum { BIO_MAX_PAGES_KMALLOC =
-               (PAGE_SIZE - sizeof(struct bio)) / sizeof(struct bio_vec),
-       MAX_PAGES_KMALLOC =
-               PAGE_SIZE / sizeof(struct page *),
-};
+enum {MAX_PAGES_KMALLOC = PAGE_SIZE / sizeof(struct page *), };
 
 unsigned exofs_max_io_pages(struct ore_layout *layout,
                            unsigned expected_pages)
@@ -49,8 +45,7 @@ unsigned exofs_max_io_pages(struct ore_layout *layout,
        unsigned pages = min_t(unsigned, expected_pages, MAX_PAGES_KMALLOC);
 
        /* TODO: easily support bio chaining */
-       pages =  min_t(unsigned, pages,
-                      layout->group_width * BIO_MAX_PAGES_KMALLOC);
+       pages =  min_t(unsigned, pages, layout->max_io_length / PAGE_SIZE);
        return pages;
 }
 

diff --git a/fs/exofs/ore.c b/fs/exofs/ore.c
index f1b7180..4ca59d4 100644 (file)
--- a/fs/exofs/ore.c
+++ b/fs/exofs/ore.c
@@ -47,9 +47,76 @@ MODULE_AUTHOR("Boaz Harrosh <bharrosh@panasas.com>");
 MODULE_DESCRIPTION("Objects Raid Engine ore.ko");
 MODULE_LICENSE("GPL");
 
+/* ore_verify_layout does a couple of things:
+ * 1. Given a minimum number of needed parameters fixes up the rest of the
+ *    members to be operatonals for the ore. The needed parameters are those
+ *    that are defined by the pnfs-objects layout STD.
+ * 2. Check to see if the current ore code actually supports these parameters
+ *    for example stripe_unit must be a multple of the system PAGE_SIZE,
+ *    and etc...
+ * 3. Cache some havily used calculations that will be needed by users.
+ */
+
 static void ore_calc_stripe_info(struct ore_layout *layout, u64 file_offset,
                                 struct ore_striping_info *si);
 
+enum { BIO_MAX_PAGES_KMALLOC =
+               (PAGE_SIZE - sizeof(struct bio)) / sizeof(struct bio_vec),};
+
+int ore_verify_layout(unsigned total_comps, struct ore_layout *layout)
+{
+       u64 stripe_length;
+
+/* FIXME: Only raid0 is supported for now. */
+       if (layout->raid_algorithm != PNFS_OSD_RAID_0) {
+               ORE_ERR("Only RAID_0 for now\n");
+               return -EINVAL;
+       }
+       if (0 != (layout->stripe_unit & ~PAGE_MASK)) {
+               ORE_ERR("Stripe Unit(0x%llx)"
+                         " must be Multples of PAGE_SIZE(0x%lx)\n",
+                         _LLU(layout->stripe_unit), PAGE_SIZE);
+               return -EINVAL;
+       }
+       if (layout->group_width) {
+               if (!layout->group_depth) {
+                       ORE_ERR("group_depth == 0 && group_width != 0\n");
+                       return -EINVAL;
+               }
+               if (total_comps < (layout->group_width * layout->mirrors_p1)) {
+                       ORE_ERR("Data Map wrong, "
+                               "numdevs=%d < group_width=%d * mirrors=%d\n",
+                               total_comps, layout->group_width,
+                               layout->mirrors_p1);
+                       return -EINVAL;
+               }
+               layout->group_count = total_comps / layout->mirrors_p1 /
+                                               layout->group_width;
+       } else {
+               if (layout->group_depth) {
+                       printk(KERN_NOTICE "Warning: group_depth ignored "
+                               "group_width == 0 && group_depth == %lld\n",
+                               _LLU(layout->group_depth));
+               }
+               layout->group_width = total_comps / layout->mirrors_p1;
+               layout->group_depth = -1;
+               layout->group_count = 1;
+       }
+
+       stripe_length = (u64)layout->group_width * layout->stripe_unit;
+       if (stripe_length >= (1ULL << 32)) {
+               ORE_ERR("Stripe_length(0x%llx) >= 32bit is not supported\n",
+                       _LLU(stripe_length));
+               return -EINVAL;
+       }
+
+       layout->max_io_length =
+               (BIO_MAX_PAGES_KMALLOC * PAGE_SIZE - layout->stripe_unit) *
+                                                       layout->group_width;
+       return 0;
+}
+EXPORT_SYMBOL(ore_verify_layout);
+
 static u8 *_ios_cred(struct ore_io_state *ios, unsigned index)
 {
        return ios->oc->comps[index & ios->oc->single_comp].cred;

diff --git a/fs/exofs/super.c b/fs/exofs/super.c
index bce3686..057b237 100644 (file)
--- a/fs/exofs/super.c
+++ b/fs/exofs/super.c
@@ -480,7 +480,7 @@ static void exofs_put_super(struct super_block *sb)
 static int _read_and_match_data_map(struct exofs_sb_info *sbi, unsigned numdevs,
                                    struct exofs_device_table *dt)
 {
-       u64 stripe_length;
+       int ret;
 
        sbi->layout.stripe_unit =
                                le64_to_cpu(dt->dt_data_map.cb_stripe_unit);
@@ -493,50 +493,7 @@ static int _read_and_match_data_map(struct exofs_sb_info *sbi, unsigned numdevs,
        sbi->layout.raid_algorithm  =
                                le32_to_cpu(dt->dt_data_map.cb_raid_algorithm);
 
-/* FIXME: Only raid0 for now. if not so, do not mount */
-       if (sbi->layout.raid_algorithm != PNFS_OSD_RAID_0) {
-               EXOFS_ERR("Only RAID_0 for now\n");
-               return -EINVAL;
-       }
-       if (numdevs < (sbi->layout.group_width * sbi->layout.mirrors_p1)) {
-               EXOFS_ERR("Data Map wrong, "
-                         "numdevs=%d < group_width=%d * mirrors=%d\n",
-                         numdevs, sbi->layout.group_width,
-                         sbi->layout.mirrors_p1);
-               return -EINVAL;
-       }
-
-       if (0 != (sbi->layout.stripe_unit & ~PAGE_MASK)) {
-               EXOFS_ERR("Stripe Unit(0x%llx)"
-                         " must be Multples of PAGE_SIZE(0x%lx)\n",
-                         _LLU(sbi->layout.stripe_unit), PAGE_SIZE);
-               return -EINVAL;
-       }
-
-       if (sbi->layout.group_width) {
-               if (!sbi->layout.group_depth) {
-                       EXOFS_ERR("group_depth == 0 && group_width != 0\n");
-                       return -EINVAL;
-               }
-               sbi->layout.group_count = numdevs / sbi->layout.mirrors_p1 /
-                                               sbi->layout.group_width;
-       } else {
-               if (sbi->layout.group_depth) {
-                       printk(KERN_NOTICE "Warning: group_depth ignored "
-                               "group_width == 0 && group_depth == %lld\n",
-                               _LLU(sbi->layout.group_depth));
-               }
-               sbi->layout.group_width = numdevs / sbi->layout.mirrors_p1;
-               sbi->layout.group_depth = -1;
-               sbi->layout.group_count = 1;
-       }
-
-       stripe_length = (u64)sbi->layout.group_width * sbi->layout.stripe_unit;
-       if (stripe_length >= (1ULL << 32)) {
-               EXOFS_ERR("Total Stripe length(0x%llx)"
-                         " >= 32bit is not supported\n", _LLU(stripe_length));
-               return -EINVAL;
-       }
+       ret = ore_verify_layout(numdevs, &sbi->layout);
 
        EXOFS_DBGMSG("exofs: layout: "
                "num_comps=%u stripe_unit=0x%x group_width=%u "
@@ -547,7 +504,7 @@ static int _read_and_match_data_map(struct exofs_sb_info *sbi, unsigned numdevs,
                _LLU(sbi->layout.group_depth),
                sbi->layout.mirrors_p1,
                sbi->layout.raid_algorithm);
-       return 0;
+       return ret;
 }
 
 static unsigned __ra_pages(struct ore_layout *layout)

diff --git a/include/scsi/osd_ore.h b/include/scsi/osd_ore.h
index 492b70d..716dbea 100644 (file)
--- a/include/scsi/osd_ore.h
+++ b/include/scsi/osd_ore.h
@@ -42,6 +42,13 @@ struct ore_layout {
        unsigned group_width;
        u64      group_depth;
        unsigned group_count;
+
+       /* Cached often needed calculations filled in by
+        * ore_verify_layout
+        */
+       unsigned long max_io_length;    /* Max length that should be passed to
+                                        * ore_get_rw_state
+                                        */
 };
 
 struct ore_dev {
@@ -138,6 +145,7 @@ static inline unsigned ore_io_state_size(unsigned numdevs)
 }
 
 /* ore.c */
+int ore_verify_layout(unsigned total_comps, struct ore_layout *layout);
 int ore_get_rw_state(struct ore_layout *layout, struct ore_components *comps,
                     bool is_reading, u64 offset, u64 length,
                     struct ore_io_state **ios);