On 12/19/2015 02:24 PM, Samuel Thibault wrote: > Hello, meta-reply > > This is another respin of IPv6 in Qemu -net user mode. > > The following patches first make some refactoring to make current code ready > for IPv6, and do not change the behavior. The actual IPv6 support will then be > submitted as a separate patch series. > > Difference with version 6 is: > - Use error_report instead of printing to stderr > - Drop extra parentheses > - Use nicer-looking sockaddr casts. > > All of this has been reviewed by Thomas Huth, thanks! It looks like you wrote your own cover letter, then supplied the git series in reply to your cover letter but with the git series still internally threaded. This resulted in the unusual reply tree of: 0/9 + 1/9 + 2/9 + 3/9 + ... + 9/9 and where clients that display trees based both on reply relationships and age would show: 0/9 + 1/9 + 2/9 | + Re: 2/9 + ... + 9/9 | + Re: 9/9 + Re: 1/9 That is, in your series, all of 2 through 9 were marked as children of 1, rather than a sibling of 1. The more typical flat threading looks like: 0/9 + 1/9 + 2/9 + ... + 9/9 where replies to 1 are then threaded closer to the original: 0/9 + 1/9 | + Re: 1/9 + 2/9 | + Re: 2/9 + ... + 9/9 + Re: 9/9 You may want to look into using 'git format-patch --cover-letter'; easily done with 'git config format.coverLetter auto'. The git cover letter would include a diffstat that includes the names of all files touched by the series as well as relative sizing of the patch, which is helpful to users. Also, using git to produce the cover letter (producing all of 0 through 9 in one shot, with 1-9 all threaded as siblings to each other and children of 0), instead of producing the body with git (just 1 through 9, with 2-9 replying to 1) and then chaining that series to a hand-written cover letter, would solve the unusual threading. > > Here is a summary of the patches: > > [PATCH 1/9] slirp: goto bad in udp_input if sosendto fails > [PATCH 2/9] slirp: Generalizing and neutralizing ARP code > [PATCH 3/9] slirp: Adding address family switch for incoming frames > [PATCH 4/9] slirp: Make Socket structure IPv6 compatible > [PATCH 5/9] slirp: Factorizing address translation > [PATCH 6/9] slirp: Factorizing and cleaning solookup() > [PATCH 7/9] slirp: Add sockaddr_equal, make solookup family-agnostic > [PATCH 8/9] slirp: Make udp_attach IPv6 compatible > [PATCH 9/9] slirp: Adding family argument to tcp_fconnect() > > -- Eric Blake eblake redhat com +1-919-301-3266 Libvirt virtualization library http://libvirt.org