December 2010 - Wine-devel - list.winehq.org

Re: winealsa.drv: use ERR() for fatal errors instead of WARN()
by Alexandre Julliard 14 Dec '10

14 Dec '10

Austin English <austinenglish(a)gmail.com> writes: > See http://bugs.winehq.org/show_bug.cgi?id=22034 If the errors are properly handled, there's no reason to have an ERR. If they aren't, that should be fixed. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH 2/2] ntdll: Check for case-insensitive volumes. (try 6 resend)
by Alexandre Julliard 14 Dec '10

14 Dec '10

Charles Davis <cdavis(a)mymail.mines.edu> writes: > +#if defined(HAVE_GETATTRLIST) && defined(ATTR_VOL_CAPABILITIES) && \ > + defined(VOL_CAPABILITIES_FORMAT) && defined(VOL_CAP_FMT_CASE_SENSITIVE) > + char *mntpoint = NULL; > + struct attrlist attr; > + struct vol_caps caps; > + struct get_fsid get_fsid; > + struct fs_cache *entry; > + > + /* First get the FS ID of the volume */ > + attr.bitmapcount = ATTR_BIT_MAP_COUNT; > + attr.reserved = 0; > + attr.commonattr = ATTR_CMN_DEVID|ATTR_CMN_FSID; > + attr.volattr = attr.dirattr = attr.fileattr = attr.forkattr = 0; > + get_fsid.size = 0; > + if (getattrlist( dir, &attr, &get_fsid, sizeof(get_fsid), 0 ) != 0 || > + get_fsid.size != sizeof(get_fsid)) > + return TRUE; > + /* Try to look it up in the cache */ > + entry = look_up_fs_cache( get_fsid.dev ); > + if (entry && !memcmp( &entry->fsid, &get_fsid.fsid, sizeof(fsid_t) )) > + /* Cache lookup succeeded */ > + return entry->case_sensitive; > + /* Cache is stale at this point, we have to update it */ > + > + mntpoint = get_device_mount_point( get_fsid.dev ); > + /* Now look up the case-sensitivity */ > + attr.commonattr = 0; > + attr.volattr = ATTR_VOL_INFO|ATTR_VOL_CAPABILITIES; > + if (getattrlist( mntpoint, &attr, &caps, sizeof(caps), 0 ) == 0) > + { If that can fail then you need to structure this differently, otherwise you'll end up looking up the mount point on every call. Also please put that chunk in a separate function, it's totally different from the statfs code path. > +#elif defined(__linux__) > + /* Only assume CIOPFS is case insensitive. */ > + /* Normally, we'd have to parse the mtab to find out exactly what > + * kind of FUSE FS this is. But, someone on wine-devel suggested > + * a shortcut. We'll stat a special file in the directory. If it's > + * there, we'll assume it's a CIOPFS, else not. > + * This will break if somebody puts a file named ".ciopfs" in a non- > + * CIOPFS directory. > + */ > + cifile = RtlAllocateHeap( GetProcessHeap(), 0, strlen( dir )+sizeof("/.ciopfs") ); > + if (!cifile) return TRUE; > + strcpy( cifile, dir ); > + strcat( cifile, "/.ciopfs" ); > + if (stat( cifile, &st ) == 0) > + { > + RtlFreeHeap( GetProcessHeap(), 0, cifile ); > + return FALSE; > + } You lost the statfs here. > @@ -1872,7 +2097,7 @@ static NTSTATUS find_file_in_dir( char *unix_name, int pos, const WCHAR *name, i > { > ret = ntdll_umbstowcs( 0, de[1].d_name, strlen(de[1].d_name), > buffer, MAX_DIR_ENTRY_LEN ); > - if (ret == length && !memicmpW( buffer, name, length)) > + if (ret == length && !memicmpW( buffer, name, length )) Please avoid gratuitous whitespace changes. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [2/2] amstream: Be more verbose when whe can't test
by Marvin 13 Dec '10

13 Dec '10

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=7579 Your paranoid android. === WVISTAADM (32 bit amstream) === Failure running script in VM: The specified guest user must be logged in interactively to perform this operation

1 0

Re: [PATCH 1/2] oleaut32: Convert ITypeInfo implementation to use standard linked list
by Alexandre Julliard 13 Dec '10

13 Dec '10

Andrew Eikum <aeikum(a)codeweavers.com> writes: > - /* search element n in list */ > - for(i=0; i < index; i++) > + i = 0; > + LIST_FOR_EACH_ENTRY (pTypeInfo, &This->typeinfo_list, ITypeInfoImpl, entry) > + { > + if(i >= index) > + break; > + ++i; > + } It looks like an array would be more appropriate than a list. -- Alexandre Julliard julliard(a)winehq.org

2 4

Removal of wineaudioio.drv
by Michael Stefaniuc 13 Dec '10

13 Dec '10

Hello guys, I have just submitted a patch to remove wineaudioio.drv from Wine. It was used by Solaris but the Solaris versions 9 and 10 that are "supported" by Wine do not have it any more. Solaris is using ESD at the moment and newer version will move to OSS4. If anybody has a reason why wineaudioio.drv should stay in please speak up *now*. This is your chance. Of course if you do speak up you volunteer to maintain wineaudioio.drv ;) bye michael

1 1

Re: ntdll: Change NtQuerySystemInformation to return 100ns ticks for SYSTEM_PROCESSOR_PERFORMANCE_INFORMATION info class (try 4)
by Chris Robinson 13 Dec '10

13 Dec '10

On Sunday, December 12, 2010 2:19:42 pm Louis Lenders wrote: > +#define USER_HZ_TO_TICKSPERSEC(x,y) (ULONGLONG)10000000 * x / y I'd recommend adding proper parenthesis, to avoid potential problems with order-of-operations. #define USER_HZ_TO_TICKSPERSEC(x,y) ((ULONGLONG)10000000 * (x) / (y)) That way it won't break if you do math in the argument for some reason, like with USER_HZ_TO_TICKSPERSEC(val+1, user_hz) or USER_HZ_TO_TICKSPERSEC(val, user_hz*2) > + static long user_hz; > + > + user_hz = sysconf(_SC_CLK_TCK); There's no real point to making it static if you're always going to (re)set the variable. It would be better to just make the call once: static long user_hz; i(user_hz == 0) user_hz = sysconf(_SC_CLK_TCK); You could also make it an inline function to avoid making user_hz a parameter: /* Convert system clock to 100ns tics */ static long user_hz; static inline ULONGLONG ClockToTics(ULONGLONG val) { i(user_hz == 0) user_hz = sysconf(_SC_CLK_TCK); return val * 10000000 / user_hz; } ... sppi->IdleTime.QuadPart = ClockToTics(idle); sppi->KernelTime.QuadPart = ClockToTics(sys); ...etc... Though I'll defer judgement to others as to what's better.

2 1

Re: msi: Fix testfailure on some 2k3 machines (try 2)
by Marvin 13 Dec '10

13 Dec '10

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=7572 Your paranoid android. === WVISTAADM (32 bit msi) === Failure running script in VM: The specified guest user must be logged in interactively to perform this operation

1 0

Re: msi: Fix testfailure on some 2k3 machines
by Hans Leidekker 13 Dec '10

13 Dec '10

On Sun, 2010-12-12 at 22:44 +0100, André Hentschel wrote: > r = MsiOpenProductA(prodcode, &hprod); > - ok(r == ERROR_INSTALL_PACKAGE_OPEN_FAILED || r == ERROR_SUCCESS, > - "Expected ERROR_INSTALL_PACKAGE_OPEN_FAILED or ERROR_SUCCESS, got %d\n", r); > + ok(r == ERROR_INSTALL_PACKAGE_OPEN_FAILED || ERROR_INSTALL_PACKAGE_REJECTED || r == ERROR_SUCCESS, > + "Expected ERROR_INSTALL_PACKAGE_OPEN_FAILED, ERROR_INSTALL_PACKAGE_REJECTED or ERROR_SUCCESS, got %d\n", r); ERROR_INSTALL_PACKAGE_REJECTED means the user has insufficient rights. Look at how action.c handles this error, for example.

2 1

Re: [PATCH] cmd: Allow GOTO to a label, which has a following whitespace, with test
by Marvin 12 Dec '10

12 Dec '10

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at http://testbot.winehq.org/JobDetails.pl?Key=7571 Your paranoid android. === WVISTAADM (32 bit) === Failure running script in VM: The specified guest user must be logged in interactively to perform this operation

1 0

Re: ntdll: Change NtQuerySystemInformation to return 100ns ticks for SYSTEM_PROCESSOR_PERFORMANCE_INFORMATION info class (try 4)
by Nikolay Sivov 12 Dec '10

12 Dec '10

On 12/13/2010 01:19, Louis Lenders wrote: > Change from try 3: improved macro as per Juan's suggestion, and made user_hz a static variable > > /proc/stat measures the times > in units of USER_HZ, whereas Windows uses 100ns ticks. > This should fix http://bugs.winehq.org/show_bug.cgi?id=23847 > @@ -1591,6 +1593,9 @@ NTSTATUS WINAPI NtQuerySystemInformation( > int count; > char name[10]; > char line[255]; > + static long user_hz; > + > + user_hz = sysconf(_SC_CLK_TCK); This change doesn't make any difference, I think the idea was to make it global.

1 0