Cc: Rafael J. Wysocki <rjw@...>, pm list <linux-pm@...>, ACPI Devel Maling List <linux-acpi@...>, Alan Stern <stern@...>, David Brownell <david-b@...>, Greg KH <greg@...>, Len Brown <lenb@...>, LKML <linux-kernel@...>, Shaohua Li <shaohua.li@...>, Alexey Starikovskiy <astarikovskiy@...>, Johannes Berg <johannes@...>
It strikes me that the only way I can think of that you could have gotten
a 7GB .git directory without really working at it is if you use one of the
so-called "dumb" git protocols that just copy whole packfiles from
kernel.org when you pull.
So do you happen to perhaps use http:// or rsync:// when you fetch git
data? That would not only be horribly slow occasionally (you'd fetch
all-new packs and re-download about 200MB of data when I repack the kernel
repo on kernel.org, which happens about once or twice every release
cycle), but it would also explain how it ballooned to 7GB for you (because
you have all these duplicate packs!).
Usign the native git:// protocol would be much faster and avoid the issue
in the future.
Linus
--