December 2015 - Wine-devel - list.winehq.org

Re: [2/2] oleaut32: Align terminating null character in SysAllocStringByteLen.
by Nikolay Sivov 25 Dec '15

25 Dec '15

On 25.12.2015 8:38, Sebastian Lackner wrote: > Signed-off-by: Sebastian Lackner <sebastian(a)fds-team.de> > --- > > If SysAllocStringByteLen() is used with an odd length, the two terminating > null bytes are not aligned properly, which means that strlenW() would not > find them. Tests show that Windows appends three terminating null bytes in > this case. > > There is no need to change the memory allocation part because we already > round up to the next bucket size. > > dlls/oleaut32/oleaut.c | 14 ++++++-------- > dlls/oleaut32/tests/vartype.c | 29 +++++++++++++++++++++++++++++ > 2 files changed, 35 insertions(+), 8 deletions(-) > > diff --git a/dlls/oleaut32/oleaut.c b/dlls/oleaut32/oleaut.c > index 75f9cb4..78cb083 100644 > --- a/dlls/oleaut32/oleaut.c > +++ b/dlls/oleaut32/oleaut.c > @@ -149,12 +149,9 @@ static bstr_t *alloc_bstr(size_t size) > > if(cache_entry) { > if(WARN_ON(heap)) { > - size_t tail; > - > - memset(ret, ARENA_INUSE_FILLER, FIELD_OFFSET(bstr_t, u.ptr[size+sizeof(WCHAR)])); > - tail = bstr_alloc_size(size) - FIELD_OFFSET(bstr_t, u.ptr[size+sizeof(WCHAR)]); > - if(tail) > - memset(ret->u.ptr+size+sizeof(WCHAR), ARENA_TAIL_FILLER, tail); > + size_t fill_size = (FIELD_OFFSET(bstr_t, u.ptr[size])+2*sizeof(WCHAR)-1) & ~(sizeof(WCHAR)-1); > + memset(ret, ARENA_INUSE_FILLER, fill_size); > + memset((char *)ret+fill_size, ARENA_TAIL_FILLER, bstr_alloc_size(size)-fill_size); > } > ret->size = size; > return ret; > @@ -418,10 +415,11 @@ BSTR WINAPI SysAllocStringByteLen(LPCSTR str, UINT len) > > if(str) { > memcpy(bstr->u.ptr, str, len); > - bstr->u.ptr[len] = bstr->u.ptr[len+1] = 0; > + bstr->u.ptr[len] = 0; > }else { > - memset(bstr->u.ptr, 0, len+sizeof(WCHAR)); > + memset(bstr->u.ptr, 0, len+1); > } > + bstr->u.str[(len+sizeof(WCHAR)-1)/sizeof(WCHAR)] = 0; > > return bstr->u.str; > } > diff --git a/dlls/oleaut32/tests/vartype.c b/dlls/oleaut32/tests/vartype.c > index 7cbb059..d905d37 100644 > --- a/dlls/oleaut32/tests/vartype.c > +++ b/dlls/oleaut32/tests/vartype.c > @@ -5444,7 +5444,9 @@ static void test_SysAllocStringByteLen(void) > { > const OLECHAR szTest[10] = { 'T','e','s','t','\0' }; > const CHAR szTestA[6] = { 'T','e','s','t','\0','?' }; > + char *buf; > BSTR str; > + int i; > > if (sizeof(void *) == 4) /* not limited to 0x80000000 on Win64 */ > { > @@ -5487,6 +5489,7 @@ static void test_SysAllocStringByteLen(void) > > ok (bstr->dwLen == 3, "Expected 3, got %d\n", bstr->dwLen); > ok (!lstrcmpA((LPCSTR)bstr->szString, szTestTruncA), "String different\n"); > + ok (!bstr->szString[2], "String not terminated\n"); > SysFreeString(str); > } > > @@ -5500,6 +5503,32 @@ static void test_SysAllocStringByteLen(void) > ok (!lstrcmpW(bstr->szString, szTest), "String different\n"); > SysFreeString(str); > } > + > + /* Make sure terminating null is aligned properly */ > + buf = HeapAlloc(GetProcessHeap(), 0, 1025); > + ok (buf != NULL, "Expected non-NULL\n"); > + for (i = 0; i < 1024; i++) > + { > + LPINTERNAL_BSTR bstr; > + > + str = SysAllocStringByteLen(NULL, i); > + ok (str != NULL, "Expected non-NULL\n"); > + bstr = Get(str); > + ok (bstr->dwLen == i, "Expected %d, got %d\n", i, bstr->dwLen); > + ok (!bstr->szString[(i+sizeof(WCHAR)-1)/sizeof(WCHAR)], "String not terminated\n"); > + SysFreeString(str); > + > + memset(buf, 0xaa, 1025); > + str = SysAllocStringByteLen(buf, i); > + ok (str != NULL, "Expected non-NULL\n"); > + bstr = Get(str); > + ok (bstr->dwLen == i, "Expected %d, got %d\n", i, bstr->dwLen); > + buf[i] = 0; > + ok (!lstrcmpA((LPCSTR)bstr->szString, buf), "String different\n"); > + ok (!bstr->szString[(i+sizeof(WCHAR)-1)/sizeof(WCHAR)], "String not terminated\n"); > + SysFreeString(str); > + } > + HeapFree(GetProcessHeap(), 0, buf); > } Does this work with disabled cache? > > static void test_SysReAllocString(void) >

4 4

Re: winedump: Add initial version of old (SLTG) typelib dumper.
by Alexandre Julliard 25 Dec '15

25 Dec '15

Dmitry Timoshkov <dmitry(a)baikal.ru> writes: > +static int _isprint(unsigned char c) > { > - int i; > + return c >= 32 ? 1 : 0; Please don't do that. > + > + assert(print_hex("magic") == 0x47544c53); Please don't do that. > + while (offset < name_table_start + name_table_size) > + { > +/*printf("current offset: %08x (relative to name table start: %08x (%u))\n", > + offset, offset - name_table_start, offset - name_table_start);*/ > + dump_binary(8); > + print_string0(); > + printf("\n"); > +/*printf("current offset: %08x (relative to name table start: %08x (%u))\n", > + offset, offset - name_table_start, offset - name_table_start);*/ Please clean these up. > +#if 0 > + print_short_hex("unknown"); > + print_short_hex("byte order mark"); > + i = tlb_read_short(); > + printf("version = %u.%u\n", i & 0xff, i >> 8); > + print_short_hex("system identifier"); > + print_hex("unknown"); > + printf("\n"); > +#else > + printf("skipping 12 bytes\n"); > + dump_binary(12); > + printf("\n"); > +#endif Same here. -- Alexandre Julliard julliard(a)winehq.org

2 2

Re: user32: Set correct caret state in the server in SetCaretPos (try 2)
by Alexandre Julliard 24 Dec '15

24 Dec '15

Anton Baskanov <baskanov(a)gmail.com> writes: > @@ -2977,7 +2977,8 @@ enum coords_relative > int x; /* caret x position */ > int y; /* caret y position */ > int hide; /* increment for hide count (can be negative to show it) */ > - int state; /* caret state (1=on, 0=off, -1=toggle current state) */ > + int state; /* caret state (1=on, 0=off, -1=toggle current state, > + * -2=on if the position differs, unchanged otherwise) */ Please define constants instead. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: server: Implement wineserver call for SystemHandleInformation. (v2)
by Marvin 24 Dec '15

24 Dec '15

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 https://testbot.winehq.org/JobDetails.pl?Key=19147 Your paranoid android. === build (build) === Patch failed to apply

2 1

Re: [2/2] server: Implement wineserver call for SystemHandleInformation.
by Alexandre Julliard 24 Dec '15

24 Dec '15

Sebastian Lackner <sebastian(a)fds-team.de> writes: > @@ -3265,6 +3265,22 @@ enum coords_relative > VARARG(sd,security_descriptor); /* retrieved security descriptor */ > @END > > + > +struct handle_info > +{ > + process_id_t owner; > + obj_handle_t handle; > + unsigned int access; > +}; > + > +/* Return a list of all opened handles */ > +(a)REQ(get_system_handles) > +(a)REPLY > + unsigned int count; /* number of handles */ > + VARARG(data,bytes); /* array of handle_infos */ This is not a bytes array, please write a proper dumping function. -- Alexandre Julliard julliard(a)winehq.org

1 0

Re: [PATCH 1/5] ole32/tests: Some tests for OleRegGetUserType()
by Marvin 24 Dec '15

24 Dec '15

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 https://testbot.winehq.org/JobDetails.pl?Key=19139 Your paranoid android. === wxppro (32 bit compobj) === compobj.c:2250: Test failed: form 0, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 1, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 2, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 3, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 4, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 0, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 1, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 2, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 3, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 4, got L"CLSID_StdFont" === w2003std (32 bit compobj) === compobj.c:2250: Test failed: form 0, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 1, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 2, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 3, got L"CLSID_StdFont" compobj.c:2250: Test failed: form 4, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 0, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 1, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 2, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 3, got L"CLSID_StdFont" compobj.c:2266: Test failed: form 4, got L"CLSID_StdFont"

1 0

Re: [PATCH] kernel32: Set error if dstlen < 0 in codepage conversion functions.
by Marvin 24 Dec '15

24 Dec '15

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 https://testbot.winehq.org/JobDetails.pl?Key=19138 Your paranoid android. === build (build) === Could not copy the patch to the VM: the "nc -q0 '10.42.42.130' '4242'" command returned 1

1 0

Re: README: add a note about 64-bit Wine and 32-bit apps (try 2)
by Alexandre Julliard 23 Dec '15

23 Dec '15

Austin English <austinenglish(a)gmail.com> writes: > @@ -80,7 +80,9 @@ Optional support libraries: > to make sure to install the 32-bit versions of these libraries; see > http://wiki.winehq.org/WineOn64bit for details. If you want a true > 64-bit Wine (or a mixed 32-bit and 64-bit Wine setup), see > - http://wiki.winehq.org/Wine64 for details. > + http://wiki.winehq.org/Wine64 for details. Note that a 64-bit Wine > + build will not run 32-bit applications (which many 64-bit applications > + include and rely upon) without also building 32-bit Wine. I think this should go on the mentioned wiki page instead, preferably with a big warning sign. -- Alexandre Julliard julliard(a)winehq.org

3 9

Re: pdh: add PdhGetLogFileTypeA/W stubs
by Sebastian Lackner 23 Dec '15

23 Dec '15

On 23.12.2015 07:36, Austin English wrote: > Fixes https://bugs.winehq.org/show_bug.cgi?id=23298 > > The function is undocumented, but from reading the documented > functions, especially PdhGetLogFileSize > (https://msdn.microsoft.com/en-us/library/windows/desktop/aa372639%28v=vs.85…) > I made an educated guess. > When there are two functions with A and W suffix, this strongly suggests that there must be a filename parameter somewhere. This means the current interpretation is definitely not correct. Did you verify if it prints useful values?

1 0

Re: [PATCH] oleaut32: Don't store non-heap-allocated strings in the BSTR cache
by Marvin 23 Dec '15

23 Dec '15

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 https://testbot.winehq.org/JobDetails.pl?Key=19120 Your paranoid android. === wxppro (32 bit vartype) === vartype: unhandled exception c0000005 at 7C929F07 === w2003std (32 bit vartype) === vartype: unhandled exception c0000005 at 7C80ED80 === wvistau64 (32 bit vartype) === vartype: unhandled exception c0000005 at 77BC1C9B vartype: unhandled exception c0000005 at 77BC25BD === w2008s64 (32 bit vartype) === vartype: unhandled exception c0000005 at 771E1C9B vartype: unhandled exception c0000005 at 771E25BD === w7u (32 bit vartype) === vartype: unhandled exception c0000005 at 77D161F6 vartype: unhandled exception c0000005 at 77D4A3CE === w7pro64 (32 bit vartype) === vartype: unhandled exception c0000005 at 770F3526 vartype: unhandled exception c0000005 at 7714AED1 === w8 (32 bit vartype) === vartype: unhandled exception c0000005 at 7712C2A7 vartype: unhandled exception c0000005 at 7719661E === w864 (32 bit vartype) === vartype: unhandled exception c0000005 at 77C00044 vartype: unhandled exception c0000005 at 77C58780 === w1064 (32 bit vartype) === vartype: unhandled exception c0000005 at 7728CCC4 vartype: unhandled exception c0000005 at 772DB260 === wvistau64 (64 bit vartype) === No test summary line found === w2008s64 (64 bit vartype) === No test summary line found === w7pro64 (64 bit vartype) === No test summary line found === w864 (64 bit vartype) === No test summary line found === w1064 (64 bit vartype) === No test summary line found

1 0