Re: [PATCH net-next v13 06/15] p4tc: add P4 data types

From
Jamal Hadi Salim <jhs@mojatatu.com>
Date
2024-03-26 23:13:15
Also in
bpf
Reply to
Paolo Abeni, 2024-03-26
On Tue, Mar 26, 2024 at 10:48 AM Paolo Abeni [off-list ref] wrote:
On Mon, 2024-03-25 at 10:28 -0400, Jamal Hadi Salim wrote:
quoted
+static int p4t_s32_validate(struct p4tc_type *container, void *value,
+                         u16 bitstart, u16 bitend,
+                         struct netlink_ext_ack *extack)
+{
+     s32 minsz = S32_MIN, maxsz = S32_MAX;
+     s32 *val = value;
+
+     if (val && (*val > maxsz || *val < minsz)) {
+             NL_SET_ERR_MSG_MOD(extack, "S32 value out of range");
I'm sorry for the additional questions/points below which could/should
have been flagged earlier.

Out of sheer ignorance IDK if a single P4 command could use multiple
types, possibly use NL_SET_ERR_MSG_FMT_MOD() and specify the bogus
value/range.
yes, that would be a more useful message. Will change.
The same point/question has a few other instances below.
quoted
+             return -EINVAL;
+     }
+
+     return 0;
+}
[...]
quoted
+static int p4t_be32_validate(struct p4tc_type *container, void *value,
+                          u16 bitstart, u16 bitend,
+                          struct netlink_ext_ack *extack)
+{
+     size_t container_maxsz = U32_MAX;
+     __be32 *val_u32 = value;
+     __u32 val = 0;
+     size_t maxval;
+     int ret;
+
+     ret = p4t_validate_bitpos(bitstart, bitend, 31, 31, extack);
+     if (ret < 0)
+             return ret;
+
+     if (value)
+             val = be32_to_cpu(*val_u32);
+
+     maxval = GENMASK(bitend, 0);
+     if (val && (val > container_maxsz || val > maxval)) {
The first condition 'val > container_maxsz' is a bit confusing
(unneeded), as 'val' type is u32 and 'container_maxsz' is U32_MAX
Good point..
quoted
+             NL_SET_ERR_MSG_MOD(extack, "BE32 value out of range");
+             return -EINVAL;
+     }
+
+     return 0;
+}
[...]
quoted
+static int p4t_u16_validate(struct p4tc_type *container, void *value,
+                         u16 bitstart, u16 bitend,
+                         struct netlink_ext_ack *extack)
+{
+     u16 container_maxsz = U16_MAX;
+     u16 *val = value;
+     u16 maxval;
+     int ret;
+
+     ret = p4t_validate_bitpos(bitstart, bitend, 15, 15, extack);
+     if (ret < 0)
+             return ret;
+
+     maxval = GENMASK(bitend, 0);
+     if (val && (*val > container_maxsz || *val > maxval)) {
Mutatis mutandis, same thing here
noted ;->
quoted
+             NL_SET_ERR_MSG_MOD(extack, "U16 value out of range");
+             return -EINVAL;
+     }
+
+     return 0;
+}
+
+static struct p4tc_type_mask_shift *
+p4t_u16_bitops(u16 bitsiz, u16 bitstart, u16 bitend,
+            struct netlink_ext_ack *extack)
+{
+     struct p4tc_type_mask_shift *mask_shift;
+     u16 mask = GENMASK(bitend, bitstart);
+     u16 *cmask;
+
+     mask_shift = kzalloc(sizeof(*mask_shift), GFP_KERNEL);
(Not specifically related to _this_ allocation) I'm wondering if the
allocations in this file should GFP_KERNEL_ACCOUNT? If I read correctly
the user-space can (and is expected to) create an quite large number of
instances of this structs (???)
I think changing to GFP_KERNEL_ACCOUNT would be useful regardless
especially because we are namespace aware.. Yes, there could be
millions of table entries in some cases.
[...]
quoted
+void __p4tc_type_host_write(const struct p4tc_type_ops *ops,
+                         struct p4tc_type *container,
+                         struct p4tc_type_mask_shift *mask_shift, void *sval,
+                         void *dval)
+{
+     #define HWRITE(cops) \
+     do { \
+             if (ops == &(cops)) \
+                     return (cops).host_write(container, mask_shift, sval, \
+                                              dval); \
+     } while (0)
+
+     HWRITE(u8_ops);
+     HWRITE(u16_ops);
+     HWRITE(u32_ops);
+     HWRITE(u64_ops);
+     HWRITE(u128_ops);
+     HWRITE(s16_ops);
+     HWRITE(s32_ops);
+     HWRITE(be16_ops);
+     HWRITE(be32_ops);
+     HWRITE(mac_ops);
+     HWRITE(ipv4_ops);
+     HWRITE(bool_ops);
+     HWRITE(dev_ops);
+     HWRITE(key_ops);
possibly

#undef HWRITE
Nod.
?

Otherwise LGTM!

Thanks for taking the time to review!

cheers,
jamal
Cheers,

Paolo
Thread (21 messages) 21 messages, 4 authors, last reply 2024-04-02
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help