Re: git: b8d60729deef - main - tcp: Congestion control cleanup.

On 11/12/21 8:19 AM, Jessica Clarke wrote:
> On 11 Nov 2021, at 11:31, Randall Stewart <rrs@FreeBSD.org> wrote:
>>
>> The branch main has been updated by rrs:
>>
>> URL: https://cgit.FreeBSD.org/src/commit/?id=b8d60729deefa0bd13e6a395fcab4928e6e10445
>>
>> commit b8d60729deefa0bd13e6a395fcab4928e6e10445
>> Author:     Randall Stewart <rrs@FreeBSD.org>
>> AuthorDate: 2021-11-11 11:28:18 +0000
>> Commit:     Randall Stewart <rrs@FreeBSD.org>
>> CommitDate: 2021-11-11 11:28:18 +0000
>>
>>     tcp: Congestion control cleanup.
>>
>>     NOTE: HEADS UP read the note below if your kernel config is not including GENERIC!!
>>
>>     This patch does a bit of cleanup on TCP congestion control modules. There were some rather
>>     interesting surprises that one could get i.e. where you use a socket option to change
>>     from one CC (say cc_cubic) to another CC (say cc_vegas) and you could in theory get
>>     a memory failure and end up on cc_newreno. This is not what one would expect. The
>>     new code fixes this by requiring a cc_data_sz() function so we can malloc with M_WAITOK
>>     and pass in to the init function preallocated memory. The CC init is expected in this
>>     case *not* to fail but if it does and a module does break the
>>     "no fail with memory given" contract we do fall back to the CC that was in place at the time.
>>
>>     This also fixes up a set of common newreno utilities that can be shared amongst other
>>     CC modules instead of the other CC modules reaching into newreno and executing
>>     what they think is a "common and understood" function. Lets put these functions in
>>     cc.c and that way we have a common place that is easily findable by future developers or
>>     bug fixers. This also allows newreno to evolve and grow support for its features i.e. ABE
>>     and HYSTART++ without having to dance through hoops for other CC modules, instead
>>     both newreno and the other modules just call into the common functions if they desire
>>     that behavior or roll there own if that makes more sense.
>>
>>     Note: This commit changes the kernel configuration!! If you are not using GENERIC in
>>     some form you must add a CC module option (one of CC_NEWRENO, CC_VEGAS, CC_CUBIC,
>>     CC_CDG, CC_CHD, CC_DCTCP, CC_HTCP, CC_HD). You can have more than one defined
>>     as well if you desire. Note that if you create a kernel configuration that does not
>>     define a congestion control module and includes INET or INET6 the kernel compile will
>>     break. Also you need to define a default, generic adds 'options CC_DEFAULT=\"newreno\"
>>     but you can specify any string that represents the name of the CC module (same names
>>     that show up in the CC module list under net.inet.tcp.cc). If you fail to add the
>>     options CC_DEFAULT in your kernel configuration the kernel build will also break.
> 
> Not doing so breaks tinderbox, as well as configs not hooks up to
> tinderbox. I don’t think this is acceptable.

We discussed this a bit on IRC, but I think in this case the default CC_*
options belong in DEFAULTS like the default GEOM_PART_* options rather
than in GENERIC.  (Though we mostly avoid changing DEFAULTS, this is one
of the rare cases when I think it makes sense.)  Handling the default for
CC_DEFAULT does not work in DEFAULTS since you can't later override it,
but that could be handled by simply having the default for CC_DEFAULT live
in the code itself under an #ifndef instead.

I think Warner is already testing a patchset to make this change.

-- 
John Baldwin