[lwip-users] Patches. Note about savannah's patch submission.

[Mike Tesch]

All,

I just submitted a number of patches for lwip.  A couple bugs
and a couple general fixup stuff.  It seems that when you submit
patches and write a comment in the Comment box, that the
comment gets thrown away (I dont see the comments I wrote
for my patches... and I dont see comments, other than the Summary,
for other people's patches.)  Just a "heads up" for people submitting
patches.

So, in lieu of that, I'll comment on the patches here.  I'm looking for
feedback on the non-bug patches.

---
netif->dhcp uninitialized - in netif_add() the field for dhcp is left
uninitialized, but is assumed to be set to NULL if dhcp doesn't ever
set it.  this causes a crash when an ARP_REPLY packet comes into
etharp_arp_input().

---
i am using dhcp/bootp to boot a system (load an OS) through tftp
so i need access to the 'si' and 'file' fields in the dhcp reply, added a
tiny bit of code to save these values.

---
there is, i believe, a bug in how memory statistics accounting in the
mem_realloc() function.  it updates lwip_stats.mem.used even if it
isn't creating a new memory chunk.  this causes mem_free() to subtract
too large of a value from mem.used when it actually frees the chunk
that mem_realloc() opted not to split up.  i'm not totally confident
this is correct, since i would think other people would have seen this
long ago; but it causes my mem.used not to shoot off to some huge
negative number when i ping flood the system.

---
i reformatted opt.h with spaces out to the same column so it's more 
readable.
hope i'm not stepping on anyone's toes here.  it's just a formatting change.

on a related note, i'm really glad to see that the more active developers
seem to have settled on a fairly consistent formatting style; it was painful
to see tabs and spaces mixed up every which way.

also, i'd like to express my appreciation for all the work y'all have 
done: thanks!

---
i also have had a bug pending in the bug system that hadn't been addressed,
so i (finally) submitted a patch for it.  it is another problem with 
mem_realloc()
where if the newsize request is unaligned, an unaligned access is generated
when ram[ptr2] is dereferenced.

---

best regards,
mt