[[!meta copyright="Copyright © 2008, 2009, 2010, 2011 Free Software Foundation, Inc."]] [[!meta license="""[[!toggle id="license" text="GFDL 1.2+"]][[!toggleable id="license" text="Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is included in the section entitled [[GNU Free Documentation License|/fdl]]."]]"""]] [[!meta title="Hiccups of Git"]] [[!tag open_issue_porting]] [[!toc]] # Log December, 2008. On the otherwise-idle flubber: $ git clone git://sources.redhat.com/git/glibc.git Initialized empty Git repository in /media/data/home/tschwinge/tmp/glibc/glibc/.git/ remote: Generating pack... remote: Done counting 380933 objects. remote: Deltifying 380933 objects... remote: 100% (380933/380933) done remote: Total 380933 (delta 294166), reused 380686 (delta 294002) Receiving objects: 100% (380933/380933), 70.31 MiB | 27 KiB/s, done. Resolving deltas: 100% (294166/294166), done. error: git-checkout-index: unable to create file iconvdata/ibm1122.c (Interrupted system call) error: git-checkout-index: unable to create file localedata/charmaps/IBM862 (Interrupted system call) Checking out files: 100% (10676/10676), done. $ git status # On branch master # Changed but not updated: # (use "git add ..." to update what will be committed) # # modified: iconvdata/ibm1122.c # modified: localedata/charmaps/IBM862 # no changes added to commit (use "git add" and/or "git commit -a") $ ls -l iconvdata/ibm1122.c localedata/charmaps/IBM862 -rw-r--r-- 1 tschwinge tschwinge 0 2008-12-15 15:49 iconvdata/ibm1122.c -rw-r--r-- 1 tschwinge tschwinge 0 2008-12-15 15:49 localedata/charmaps/IBM862 So these files are indeed of zero-length in the checked-out tree. Is this Git's fault or something else's? Fixing this situation is easy enough: $ git checkout -- iconvdata/ibm1122.c localedata/charmaps/IBM862 $ git status # On branch master nothing to commit (working directory clean) Still seen on 2010-03-16. --- A very similar issue, seen on 2010-11-17. The working tree had a lot of differences to HEAD. tschwinge@grubber:~/tmp/gcc/hurd $ git reset --hard HEAD error: unable to unlink old 'gcc/config/darwin.h' (Interrupted system call) Checking out files: 100% (1149/1149), done. fatal: Could not reset index file to revision 'HEAD'. tschwinge@grubber:~/tmp/gcc/hurd $ git reset --hard HEAD error: unable to unlink old 'gcc/config/iq2000/iq2000.md' (Interrupted system call) error: git checkout-index: unable to create file gcc/config/lm32/lm32.c (File exists) Checking out files: 100% (1149/1149), done. fatal: Could not reset index file to revision 'HEAD'. tschwinge@grubber:~/tmp/gcc/hurd $ ls -l gcc/config/iq2000/iq2000.md gcc/config/lm32/lm32.c ls: cannot access gcc/config/iq2000/iq2000.md: No such file or directory -rw-r--r-- 1 tschwinge tschwinge 32159 Nov 17 19:09 gcc/config/lm32/lm32.c tschwinge@grubber:~/tmp/gcc/hurd $ git reset --hard HEAD error: git checkout-index: unable to create file gcc/fortran/expr.c (Interrupted system call) Checking out files: 100% (1149/1149), done. fatal: Could not reset index file to revision 'HEAD'. tschwinge@grubber:~/tmp/gcc/hurd $ git reset --hard HEAD error: git checkout-index: unable to create file gcc/config/sol2.h (Interrupted system call) Checking out files: 100% (1149/1149), done. fatal: Could not reset index file to revision 'HEAD'. tschwinge@grubber:~/tmp/gcc/hurd $ git reset --hard HEAD error: unable to unlink old 'gcc/config/i386/i386.c' (Interrupted system call) Checking out files: 100% (1149/1149), done. fatal: Could not reset index file to revision 'HEAD'. tschwinge@grubber:~/tmp/gcc/hurd $ git reset --hard HEAD Checking out files: 100% (1149/1149), done. HEAD is now at fe3e43c Merge commit 'refs/top-bases/hurd/master' into hurd/master --- 2010-12-22, grubber: $ git remote update Fetching savannah remote: Counting objects: 582331, done. remote: Compressing objects: 100% (124133/124133), done. remote: Total 582331 (delta 460856), reused 578352 (delta 457598) Receiving objects: 100% (582331/582331), 525.15 MiB | 204 KiB/s, done. fatal: cannot pread pack file: Interrupted system call fatal: index-pack failed error: Could not fetch savannah --- 2011-06-10, coulomb.SCHWINGE, checking out [[binutils]]' master branch, starting from an empty working directory (after an external `git push`): $ git checkout -f fatal: cannot create directory at 'gas/testsuite/gas/bfin': Interrupted system call $ git checkout -f error: unable to create file gas/testsuite/gas/i386/ilp32/x86-64-sse4_1-intel.d (File exists) warning: unable to unlink gas/testsuite/gas/m68k-coff: Operation not permitted fatal: cannot create directory at 'gas/testsuite/gas/m68k-coff': Operation not permitted $ git checkout -f error: unable to create file gas/testsuite/gas/h8300/h8300.exp (File exists) error: unable to create file gas/testsuite/gas/i386/x86-64-addr32-intel.d (File exists) error: unable to create file gas/testsuite/gas/ia64/secname.d (File exists) error: unable to create file gas/testsuite/gas/m68k/pr11676.s (File exists) Checking out files: 100% (12315/12315), done. $ git status # On branch master # Changes not staged for commit: # (use "git add ..." to update what will be committed) # (use "git checkout -- ..." to discard changes in working directory) # # modified: gas/testsuite/gas/h8300/h8300.exp # modified: gas/testsuite/gas/i386/x86-64-addr32-intel.d # modified: gas/testsuite/gas/ia64/secname.d # modified: gas/testsuite/gas/m68k/pr11676.s # no changes added to commit (use "git add" and/or "git commit -a") $ rm gas/testsuite/gas/h8300/h8300.exp gas/testsuite/gas/i386/x86-64-addr32-intel.d gas/testsuite/gas/ia64/secname.d gas/testsuite/gas/m68k/pr11676.s $ git checkout -f $ git status # On branch master nothing to commit (working directory clean) # Analysis 2011-06-13 Running `git checkout -f` under GDB: error: git checkout-index: unable to create file gas/testsuite/gas/cris/string-1.s (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/i386/x86-64-sse-check.d (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/i386/x86-64-sse4_1.d (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/ppc/astest.d (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/tic6x/reloc-bad-4.s (File exists) warning: unable to unlink include/cgen: Operation not permitted fatal: cannot create directory at 'include/cgen': Operation not permitted Again: error: git checkout-index: unable to create file gas/config/te-vxworks.h (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/cris/string-1.s (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/d10v/warning-019.s (File exists) error: git checkout-index: unable to create file gas/testsuite/gas/i860/dual03.s (File exists) error: git checkout-index: unable to create file ld/testsuite/ld-mmix/sec-7a.s (File exists) warning: unable to unlink ld/testsuite/ld-powerpc: Operation not permitted fatal: cannot create directory at 'ld/testsuite/ld-powerpc': Operation not permitted And: [[git_duplicated_content]]. All these (very likely) have the same root cause: `SA_RESTART` restarting too much. With `git checkout`, Git uses in progress.c a SIGALRM handler (`SA_RESTART`; invoked every second via `setitimer(ITIMER_REAL)`) to display status messages: *x % already checked out*. To avoid the status update signals every second, in `[git]/progress.c:start_progress_delay` we can just return `NULL` (manually in GDB, for example), then both the *error: git checkout-index* and the [[duplicated content|git_duplicated_content]] issues go away. I'm guessing that when returning from a `SA_RESTART` signal handler, too much of the \`\`syscall''s is being restarted. For example, if a file has already been created, the restarted creation attempt would fail: *File exists*. If data has been written, it might get written again (duplication issue). Then, there are cases where `unlink` apparently returns EINTR, which is not kosher either. Etc. Do we have problems with `SA_RESTART` vs. the atomicity of our syscall-alikes?