c-deps: bump jemalloc version

[nicktrav]

Bump the version of jemalloc from jemalloc/jemalloc@54eaed1 (5.3.0) to jemalloc/jemalloc@17881eb (tip of dev at time of commit).

Commit log of changes can be generated with the following:

$ cd ./cdeps/jemalloc
$ git log --pretty='format:%C(auto)%h (%as) %s' 54eaed1..17881eb

The full diff of changes can be found here.

Release note (general change): The version of jemalloc (the memory allocator that CRDB uses) to the latest version. This pulls in a number of upstream fixes and improvements.

Epic: None.

[cockroach-teamcity]

This change is 

[nicktrav]

@rickystewart - would you be able to weave your magic here and generate the c-deps?

[rickystewart]

Give me an hour or so.

[rickystewart]

Is 17881ebbfd76529904e826f425f3266834cf3a75 not pushed? The build is failing.

[nicktrav]

Should be good now. I see the rev there on the dev branch of our fork.

[rickystewart]

OK, I've kicked off the build, please wait until Tuesday and I'll send you a patch.

[rickystewart]

Looks like the build fails now:

23:58:31 src/prof_sys.c:143:1: error: unknown warning group '-Wframe-address', ignored [-Werror,-Wunknown-warning-option]
23:58:31 JEMALLOC_DIAGNOSTIC_IGNORE_FRAME_ADDRESS
23:58:31 ^
23:58:31 include/jemalloc/internal/jemalloc_internal_macros.h:87:6: note: expanded from macro 'JEMALLOC_DIAGNOSTIC_IGNORE_FRAME_ADDRESS'
23:58:31 JEMALLOC_DIAGNOSTIC_IGNORE("-Wframe-address")
23:58:31 ^
23:58:31 include/jemalloc/internal/jemalloc_internal_macros.h:71:6: note: expanded from macro 'JEMALLOC_DIAGNOSTIC_IGNORE'
23:58:31 JEMALLOC_PRAGMA__(GCC diagnostic ignored W)
23:58:31 ^
23:58:31 include/jemalloc/internal/jemalloc_internal_macros.h:67:32: note: expanded from macro 'JEMALLOC_PRAGMA__'
23:58:31 # define JEMALLOC_PRAGMA__(X) _Pragma(#X)
23:58:31 ^
23:58:31 :123:25: note: expanded from here
23:58:31 GCC diagnostic ignored "-Wframe-address"

[RaduBerinde]

I am a bit hesitant to switch to something that doesn't have a release. Part of it is that probably not many other jemalloc users are using the dev branch so we may be the first to hit an issue.
It's a tradeoff so we should figure out if any of the new commits are a significant improvement that we really want.

[nicktrav]

@RaduBerinde - sounds good. Let me at least get this building, so we can take it for a spin, and we can make a call then. Looks like there's some breakage that we'll need to understand / fix anyway.