Re: [RFC/PATCH 0/1] cygwin: Remove the Win32 l/stat() functions
From: Torsten Bögershausen <hidden>
Date: 2016-06-15 22:57:55
On 25.06.13 21:23, Johannes Sixt wrote:
Am 25.06.2013 00:10, schrieb Junio C Hamano:quoted
Mark Levedahl [off-list ref] writes:quoted
On 06/22/2013 03:38 PM, Ramsay Jones wrote:quoted
Also, apart from running the git test-suite, I have the Win32 l/stat functions disabled on all of my repos. In particular, I have core.filemode set to true. (At one point, I used to build git with NO_TRUSTABLE_FILEMODE reset so that I wouldn't have to remember to reset core.filemode by hand after a git-clone or git-init). I should also note that I run MinGW git on the same laptop and, using git.git as an example, it does not seem that much faster (if at all) than cygwin git.After applying your patch to master, I've had the test-suite running in a VM using WinXP + current cygwin (v1.7.x) for about 8 hours, no failures so far but it could take another day to complete. I never found any real speed up using the Win32 stat/lstat functions, and the lack of Posix compatibility breaking cross-platform projects, links, etc., made this function a mis-feature in my opinion for years. As I found no positive benefit from the Win32 lstat, I modified git for local use years ago to set core.filemode=true when cloning / initing to avoid as many issues as possible.So that's two votes to use the vanilla Cygwin stat/lstat, essentially reverting adbc0b6b (cygwin: Use native Win32 API for stat, 2008-09-30), which was added by Dmitry and Shawn while I was away. Let's wait and see if people give us more data points and decide.That'll be more productive if we let the list know ;-) Some context: This is about a patch by Ramsay that removes the "schizophrenic lstat" hack for Cygwin. Junio, can you please queue that patch in pu? -- Hannes
Here some benchmarks: Clone the linux kernel from August 2012 (which was "in house") to the windows box, 2.3 GhZ Core duo. Run with and without core.filemode, which triggers cygwinfstricks (and is shorter to type) Numbers are typical "hot cache", cold cache is 6..8 seconds "real" $ git --version git version 1.8.3.1.g6f17ca7 __________________________________ $ time git -c core.filemode=true status -uno # On branch master # Changes not staged for commit: # (use "git add <file>..." to update what will be committed) # (use "git checkout -- <file>..." to discard changes in working directory) # # modified: include/linux/netfilter/xt_CONNMARK.h # modified: include/linux/netfilter/xt_DSCP.h # modified: include/linux/netfilter/xt_MARK.h # modified: include/linux/netfilter/xt_RATEEST.h # modified: include/linux/netfilter/xt_TCPMSS.h # modified: include/linux/netfilter_ipv4/ipt_ECN.h # modified: include/linux/netfilter_ipv4/ipt_TTL.h # modified: include/linux/netfilter_ipv6/ip6t_HL.h # modified: net/netfilter/xt_DSCP.c # modified: net/netfilter/xt_HL.c # modified: net/netfilter/xt_RATEEST.c # modified: net/netfilter/xt_TCPMSS.c # no changes added to commit (use "git add" and/or "git commit -a") real 0m2.313s user 0m0.577s sys 0m1.765s ------------------------------------------------------ tb@snoopy ~/projects/linux-2.6 $ time git -c core.filemode=false status -uno # On branch master # Changes not staged for commit: # (use "git add <file>..." to update what will be committed) # (use "git checkout -- <file>..." to discard changes in working directory) # # modified: include/linux/netfilter/xt_CONNMARK.h # modified: include/linux/netfilter/xt_DSCP.h # modified: include/linux/netfilter/xt_MARK.h # modified: include/linux/netfilter/xt_RATEEST.h # modified: include/linux/netfilter/xt_TCPMSS.h # modified: include/linux/netfilter_ipv4/ipt_ECN.h # modified: include/linux/netfilter_ipv4/ipt_TTL.h # modified: include/linux/netfilter_ipv6/ip6t_HL.h # modified: net/netfilter/xt_DSCP.c # modified: net/netfilter/xt_HL.c # modified: net/netfilter/xt_RATEEST.c # modified: net/netfilter/xt_TCPMSS.c # no changes added to commit (use "git add" and/or "git commit -a") real 0m1.047s user 0m0.311s sys 0m0.765s /Torsten