observation is that some 570x devices can get themselves into a state
where they miscompute off-loaded TCP or UDP checksums on packets so
small that Ethernet padding is required. Further obsevation suggests
that the bge checksum-offload hardware is adding those padding bytes
into its TCP checksum computation. (Once a 5700 gets in this state,
even a warm boot won't fix it: it needs a hard powerdown.)
Work around the problem by padding such runts with zeros: even if the
checksum-offload adds in extra zeros, the resulting sum will be correct.
Also, dont trust the checksum-offload on received packets smaller than
the minimum ethernet frame, in case the Rx-side has a similar bug.
Finally, on packets where we do trust the outboard Rx-side TCP or UDP
checksum, the bge did not include the pseudo-header. Set the
M_CSUM_NO_PSEUDOHDR bit as well as M_CSUM_DATA, and rely on
udp_input() or tcp_input() adding in the sum via in_cksum_phdr().
* use ${G_HCONFIG_H} instead of hconfig.h as a dependency.
* hconfig.h: auto-build.h dependency is bogus, remove.
* misc cleanup of ".c depends on .h" -> ".o depends on .h"
* options.h was missing from CLEANFILES.
With these changes running make in this directory doesn't do a
*complete* rebuild anymore. Still not foolproof, because gen* host
tools do not overwrite generated files that are not changed, so we
really need to use stamp file style dependencies for those, like gcc
makefile does.
Ok by Matthew Green <mrg@netbsd>.
if we m_dup() a packet to compactify it, and later run out of DMA
descriptors, bge_encap() will return ENOBUFS, hoping the driver will
try again later. But we have just m_freem()'d the original chain
which was m_dup()'d, leaving a pointer to the just-freed packet header
in the tx queue.
Fix by always walking the chain, shuffling data towards the head;
except if we find a runt in the very last mbuf, we must borrow data
from its predecessor.
(Patch is verbatim from a third-party tree, apologies for any style woes.)
Sam Leffler's FreeBSD commit message was
``to eliminate context switch when returning results from the
software crypto driver''
but the patch also contains the CRYPTO_SESID*() macros used in newer
ubsec and hifn drivers.
-c used to specify the number of cylinders per group
- but newfs has ignored cylinders for a while.
Changing what -c meant is just asking for trouble!
I might add something to let MINCYL be changed.
- allows less than 'one fragment per inode' (useful for mfs /dev)
- limits number of inodes to 2^31 (they are stored in an int32_t)
- errors if the number of cylinder groups is such that the cylinder group
summary won't fit in the first cylinder group.
- ensures that the last cylinder block contains a valid number of fragments
and inodes, and is not larger than any earlier ones.
- cylinder groups are now created with almost the same size as each other.
Change posted to tech-kern, and no one objected.
itojun