Re: GetVolumeInformationA: please test this on MS-Windows

[Andrew Eikum]

On 07/29/2010 03:43 AM, [e-mail] wrote:
> Andrew,
> > Run on a Win98SE VM.  The E drive is a CDROM drive.
>
> Thank you very much for the result. I added the broken(GetLastError()) for Win9x
> based on your data.
>
> Now is your chance to repeat the test with
> https://testbot.winehq.org/JobDetails.pl?Key=4014
> (either as .patch file or the binary from testbot).
> This is valuable because testbot skips most new tests, lacking another drive,
> giving no proof that the fixed patch indeed works without errors on win9x.
> If it fails, please report to wine-devel as I already submitted the new patch.

Hi Jörg,

No failures with the testbot executable.  Run under the same Win98SE VM, 
again from the E: drive.

Andrew

GetVolumeInformationA: please test this on MS-Windows

[Jeff Zaroyko]

On Thu, Jul 29, 2010 at 11:17 PM, Andrew Eikum  wrote:
> On 07/29/2010 03:43 AM, [e-mail] wrote:
> > Andrew,
> > > Run on a Win98SE VM.  The E drive is a CDROM drive.
> >
> > Thank you very much for the result. I added the broken(GetLastError()) for
> > Win9x
> > based on your data.
> >
> > Now is your chance to repeat the test with
> > https://testbot.winehq.org/JobDetails.pl?Key=4014
> > (either as .patch file or the binary from testbot).
> > This is valuable because testbot skips most new tests, lacking another
> > drive,
> > giving no proof that the fixed patch indeed works without errors on win9x.
> > If it fails, please report to wine-devel as I already submitted the new
> > patch.
>
> Hi Jörg,
>
> No failures with the testbot executable.  Run under the same Win98SE VM,
> again from the E: drive.
>
> Andrew

Out of curiosity, I ran your test binary in my MinGW shell  (as part
of msysgit)  under Windows Vista SP2 32bit.  Regular winetest doesn't
have any additional kernel32 volume test failures, but yours:

$ kernel32_test.exe volume
volume.c:395: Test failed: GetVolumeInformationA failed, last error 123
volume.c:406: Test failed: GetVolumeInformationA failed, last error 123
volume: 405 tests executed (0 marked as todo, 2 failures), 0 skipped.

$ uname -a
MINGW32_NT-6.0 JEFFZ-PC 1.0.11(0.46/3/2) 2009-02-27 22:34 i686 unknown

jeffz@JEFFZ-PC ~/Desktop
$ pwd
/c/Users/jeffz/Desktop

Regards
Jeff

GetVolumeInformationA: please test this on MS-Windows

Hi,

The binary from https://testbot.winehq.org/JobDetails.pl?Key=4014
passes all tests on native
win95
w2k SP4
wxp SR3
and inside vmware on
win98SE (Andrew Eikum)

Jeff Zaroyko writes:
> I ran your test binary in my MinGW shell  (as part
> of msysgit)  under Windows Vista SP2 32bit.

I can imagine two possible reasons why it fails on your Vista machine
a) Testbot's mingw compiled binary running in the mingw shell may
   behave differently than a native bin in a native shell.

b) SetEnvironmentVariable("=C:",path) is not usable on/since(?) Vista
   to modify the per-drive default directory and hence does not manage
   to establish the precondition needed for the test.

Do you get different results when doing the following:
# assuming C: is where %WINDOWSDIR% resides
C:
cd \
X:    # which dir on X should not matter
Y:\path\to\kernel32_test.exe volume
C:
cd \windows
X:
Y:\path\to\kernel32_test.exe volume
# X: must not be C:; Y: can be any drive

For completion, you could check whether you get different results when
invoking the test.exe with current directory
a) C:\          # many GetVolumeInfo tests are skipped
b) C:\windows   # dito
c) X:\          with \windows as CD on C:
d) X:\anysubdir with \windows as CD on C:
e) X:\anysubdir with \ as CD on C:
f) X:\          with \ as CD on C:
I think that if SetEnv has no effect you'll get different failures.

I.e. this batch file
C:
CD \
Y:\path\to\kernel32_test.exe volume
CD windows 
Y:\path\to\kernel32_test.exe volume
X:
CD \
Y:\path\to\kernel32_test.exe volume
CD \anysubdir
Y:\path\to\kernel32_test.exe volume
C:
CD \
X:
Y:\path\to\kernel32_test.exe volume
CD \
Y:\path\to\kernel32_test.exe volume

Thank you,
    Jörg Höhle

GetVolumeInformationA: please test this on MS-Windows

[Jeff Zaroyko]

On Fri, Jul 30, 2010 at 7:01 PM,   wrote:
> Do you get different results when doing the following:
> # assuming C: is where %WINDOWSDIR% resides

There's a %WINDIR% which is c:\windows
> C:
> cd \
> X:    # which dir on X should not matter
> Y:\path\to\kernel32_test.exe volume

Passes outside of MinGW.
> C:
> cd \windows
> X:
> Y:\path\to\kernel32_test.exe volume
> # X: must not be C:; Y: can be any drive

Passes outside of MinGW.
> For completion, you could check whether you get different results when
> invoking the test.exe with current directory
> a) C:\          # many GetVolumeInfo tests are skipped
> b) C:\windows   # dito
> c) X:\          with \windows as CD on C:
> d) X:\anysubdir with \windows as CD on C:
> e) X:\anysubdir with \ as CD on C:
> f) X:\          with \ as CD on C:
> I think that if SetEnv has no effect you'll get different failures.
>
> I.e. this batch file
> C:
> CD \
> Y:\path\to\kernel32_test.exe volume
> CD windows
> Y:\path\to\kernel32_test.exe volume
> X:
> CD \
> Y:\path\to\kernel32_test.exe volume
> CD \anysubdir
> Y:\path\to\kernel32_test.exe volume
> C:
> CD \
> X:
> Y:\path\to\kernel32_test.exe volume
> CD \
> Y:\path\to\kernel32_test.exe volume

No failures outside of MinGW.
Inside MinGW, I invoke cmd to run the batch file:

jeffz@JEFFZ-PC ~/Desktop
$ cmd
Microsoft Windows [Version 6.0.6002]
Copyright (c) 2006 Microsoft Corporation.  All rights reserved.

c:\Users\jeffz\Desktop>joerg.bat

c:\Users\jeffz\Desktop>C:

c:\Users\jeffz\Desktop>CD \

c:>c:\users\jeffz\Desktop\kernel32_test.exe volume
volume.c:365: Test failed: GetVolumeInformationA did not fail, last error 373
8559
volume.c:384: Test failed: GetVolumeInformationA did not fail, last error 373
8559
volume: 406 tests executed (0 marked as todo, 2 failures), 0 skipped.

c:>CD windows

c:\Windows>c:\users\jeffz\Desktop\kernel32_test.exe volume
volume.c:395: Test failed: GetVolumeInformationA failed, last error 123
volume.c:406: Test failed: GetVolumeInformationA failed, last error 123
volume: 406 tests executed (0 marked as todo, 2 failures), 0 skipped.

c:\Windows>D:

D:>CD \

D:>c:\users\jeffz\Desktop\kernel32_test.exe volume
volume: 406 tests executed (0 marked as todo, 0 failures), 0 skipped.

D:>CD \recovery

D:\RECOVERY>c:\users\jeffz\Desktop\kernel32_test.exe volume
volume: 406 tests executed (0 marked as todo, 0 failures), 0 skipped.

D:\RECOVERY>C:

c:\Windows>CD \

c:>D:

D:\RECOVERY>c:\users\jeffz\Desktop\kernel32_test.exe volume
volume: 406 tests executed (0 marked as todo, 0 failures), 0 skipped.

D:\RECOVERY>CD \

D:>c:\users\jeffz\Desktop\kernel32_test.exe volume
volume: 406 tests executed (0 marked as todo, 0 failures), 0 skipped.

D:>

Regards
Jeff

GetVolumeInformationA: please test this on MS-Windows

Hi,

Jeff Zaroyko wrote:
> No failures outside of MinGW.

Thanks. This validates the relationship between GetVolumeInformation
and the per-drive current directory on win95/98/2k/xp/7.
> Inside MinGW, I invoke cmd to run the batch file:
> volume: 406 tests executed (0 marked as todo, 2 failures), 0 skipped.

Wait! When starting from C: as in cases a) and b) there should have been a skip.
    if (currentdir[0] == windowsdir[0]) {
        skip("Please re-run from another device than %c:\n", windowsdir[0]);
If properly skipped, there would have been 0 failures.

Could you please add a
TRACE("%s vs. %s\n", currentdir, windowsdir);
Either the path is in a strange format, or there's a case issue.
Perhaps GetCurrentDir reports "C:..." while GetWindowsDir uses "c:..."?
Then I should use toupper(currentdir[0]) == toupper(windowsdir[0])
Could you please retry the test with
if (toupper(currentdir[0]) == toupper(windowsdir[0])) {
    skip(...)
> a) C:\          # many GetVolumeInfo tests are skipped
> b) C:\windows   # dito
> c) X:\          with \windows as CD on C:
> d) X:\anysubdir with \windows as CD on C:
> e) X:\anysubdir with \ as CD on C:
> f) X:\          with \ as CD on C:

Regards,
 Jörg Höhle

GetVolumeInformationA: please test this on MS-Windows

[Jeff Zaroyko]

On Mon, Aug 2, 2010 at 7:03 PM,   wrote:
> Hi,
>
> Jeff Zaroyko wrote:
> > No failures outside of MinGW.
>
> Thanks. This validates the relationship between GetVolumeInformation
> and the per-drive current directory on win95/98/2k/xp/7.
> > Inside MinGW, I invoke cmd to run the batch file:
> > volume: 406 tests executed (0 marked as todo, 2 failures), 0 skipped.
>
> Wait! When starting from C: as in cases a) and b) there should have been a skip.
>    if (currentdir[0] == windowsdir[0]) {
>        skip("Please re-run from another device than %c:\n", windowsdir[0]);
> If properly skipped, there would have been 0 failures.
>
> Could you please add a
> TRACE("%s vs. %s\n", currentdir, windowsdir);

Would you mind supplying a new binary?  My regular GNU/Linux system is
dead due to hardware issues and I don't have a new enough version of
flex to try building with MinGW on Windows.

GetVolumeInformationA: please test this on MS-Windows

Jeff Zaroyko wrote:
> Would you mind supplying a new binary?

Testbot did it for us: job #4154
https://testbot.winehq.org/JobDetails.pl?Key=4154

Regards,
 Jörg Höhle

GetVolumeInformationA: please test this on MS-Windows

[Jeff Zaroyko]

On Mon, Aug 2, 2010 at 9:43 PM,   wrote:
> Jeff Zaroyko wrote:
> > Would you mind supplying a new binary?
>
> Testbot did it for us: job #4154
> https://testbot.winehq.org/JobDetails.pl?Key=4154
>
> Regards,
>  Jörg Höhle

This new #4154 binary doesn't have any failures when run from the MinGW shell:

volume.c:345: Tests skipped: Please re-run from another device than C:
volume: 392 tests executed (0 marked as todo, 0 failures), 1 skipped.

Only the earlier binary from #4014 does as per the earlier email.

Jeff