July 2022 - Wine-Devel - list.winehq.org

Re: [PATCH v2 3/5] mshtml: Implement document.doctype.

by Marvin

Hi, It looks like your patch introduced the new failures shown below. Please investigate and fix them before resubmitting your patch. If they are not new, fixing them anyway would help a lot. Otherwise please ask for the known failures list to be updated. The tests also ran into some preexisting test failures. If you know how to fix them that would be helpful. See the TestBot job for the details: The full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=120190 Your paranoid android. === w10pro64_en_AE_u8 (64 bit report) === mshtml: htmldoc.c:2541: Test failed: unexpected call UpdateUI htmldoc.c:2853: Test failed: unexpected call Exec_UPDATECOMMANDS htmldoc.c:350: Test failed: expected Exec_SETTITLE htmldoc.c:2859: Test failed: unexpected call Exec_SETTITLE

July 29, 2022

Re: [PATCH 4/5] mshtml: Implement DocumentType node's name prop.

by Marvin

Hi, It looks like your patch introduced the new failures shown below. Please investigate and fix them before resubmitting your patch. If they are not new, fixing them anyway would help a lot. Otherwise please ask for the known failures list to be updated. The tests also ran into some preexisting test failures. If you know how to fix them that would be helpful. See the TestBot job for the details: The full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=120170 Your paranoid android. === w10pro64_ja (64 bit report) === mshtml: htmldoc.c:2541: Test failed: unexpected call UpdateUI htmldoc.c:2853: Test failed: unexpected call Exec_UPDATECOMMANDS htmldoc.c:2541: Test failed: unexpected call UpdateUI htmldoc.c:2853: Test failed: unexpected call Exec_UPDATECOMMANDS htmldoc.c:350: Test failed: expected Exec_SETTITLE htmldoc.c:2859: Test failed: unexpected call Exec_SETTITLE

July 29, 2022

Re: [PATCH 3/5] mshtml: Implement document.doctype.

by Marvin

Hi, It looks like your patch introduced the new failures shown below. Please investigate and fix them before resubmitting your patch. If they are not new, fixing them anyway would help a lot. Otherwise please ask for the known failures list to be updated. The tests also ran into some preexisting test failures. If you know how to fix them that would be helpful. See the TestBot job for the details: The full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=120169 Your paranoid android. === w1064v1809 (testbot log) === WineRunTask.pl:error: The previous 1 run(s) terminated abnormally

July 29, 2022

Re: [PATCH vkd3d 2/3] vkd3d-shader: Introduce a compile option to use Unknown format for typed UAV reads.

by Conor McCarthy

On Fri, Jul 22, 2022 at 8:28 PM Henri Verbeet <hverbeet(a)gmail.com> wrote: > I'm not entirely sure we're decoding these correctly. The format is > certainly specified in the HLSL; it would seem surprising (though not > inconceivable) that the information just gets dropped when generating > the bytecode. > It appears from the spec we are not missing anything. The obvious way to encode a format component count is in the dword which specifies the type of each component. Those beyond the last component could be D3D11_SB_RETURN_TYPE_UNUSED, but instead they all have the same type. There's no other place to store it. The output from /dumpbin has four components for the declaration too, and the load instruction specifies a vec4 load. Looks like the actual format component count is left out.

July 29, 2022

[PATCH v2] ntdll/signalx86_64: Add CFI for __wine_syscall_dispatcher

by Keno Fischer

In a debugger that is capable of loading both PE and ELF libraries (e.g. the wine-enhanced GDB version at [1]), this unwind information will allow proper unwind across the __wine_syscall_dispatcher. The CFI itself is relatively straightforward, though the implementation looks a little complex, because support for proper CFI encoding is somewhat limitied in the GNU toolchain. Essentially, as registers are stored in the rcx buffer in this function, we switch their cfi description to point at that rcx buffer, so the registers may be freely used thereafter without affecting the unwind. Eventually the rcx buffer is tracked in rbp instead at which point we pivot the CFI information as well and keep it there for most of the rest of the function. [1] https://github.com/JuliaComputing/gdb-solib-wine Signed-off-by: Keno Fischer <keno(a)juliacomputing.com> --- v2: Fix accidental build issues introduced during cleanup. (And missed in my build test, because I forgot to enable win64, sigh). dlls/ntdll/unix/signal_x86_64.c | 181 ++++++++++++++++++++++++++++++++ 1 file changed, 181 insertions(+) diff --git a/dlls/ntdll/unix/signal_x86_64.c b/dlls/ntdll/unix/signal_x86_64.c index d6f19dbfa84..2d33df8bccd 100644 --- a/dlls/ntdll/unix/signal_x86_64.c +++ b/dlls/ntdll/unix/signal_x86_64.c @@ -442,6 +442,10 @@ static USHORT fs32_sel; /* selector for %fs in 32-bit mode */ * Definitions for Dwarf unwind tables */ +/* For use in inline-assembly and other contexts where C enums are unavailable */ +#define PP_DW_CFA_def_cfa_expression 0xf +#define PP_DW_CFA_expression 0x10 + enum dwarf_call_frame_info { DW_CFA_advance_loc = 0x40, @@ -472,6 +476,14 @@ enum dwarf_call_frame_info DW_CFA_val_expression = 0x16, }; +/* DWARF State machine ops */ +#define PP_DW_OP_const1u 0x08 +#define PP_DW_OP_breg_base 0x70 +#define PP_DW_OP_minus 0x1c +#define PP_DW_OP_deref 0x06 +#define PP_DW_OP_plus 0x22 +#define PP_DW_OP_plus_uconst 0x23 + enum dwarf_operation { DW_OP_addr = 0x03, @@ -3288,35 +3300,177 @@ __ASM_GLOBAL_FUNC( signal_exit_thread, __ASM_CFI(".cfi_rel_offset %r15,8\n\t") "call *%rsi" ) + +/*********************************************************************** + * CFI Utility definitions + * + * These utility definitions help encode CFI unwind information (for consumption + * in gdb and similar debuggers), improving the accuracy of backtraces. + */ + +/* DWARF Register ordering for x86_64 */ +#define X86_64_DWARF_RAX 0x00 +#define X86_64_DWARF_RDX 0x01 +#define X86_64_DWARF_RCX 0x02 +#define X86_64_DWARF_RBX 0x03 +#define X86_64_DWARF_RSI 0x04 +#define X86_64_DWARF_RDI 0x05 +#define X86_64_DWARF_RBP 0x06 +#define X86_64_DWARF_RSP 0x07 +#define X86_64_DWARF_R8 0x08 +#define X86_64_DWARF_R9 0x09 +#define X86_64_DWARF_R10 0x0a +#define X86_64_DWARF_R11 0x0b +#define X86_64_DWARF_R12 0x0c +#define X86_64_DWARF_R13 0x0d +#define X86_64_DWARF_R14 0x0e +#define X86_64_DWARF_R15 0x0f +#define X86_64_DWARF_RIP 0x10 + +/* __ASM_REG_IS_DEREF_REF_OFFSET(REG_SAVE, REG, OFFSET) + * + * Declare that the register with DWARF register number `REG` is + * saved at offset `OFFSET` from the register with DWARF register + * number `REG_SAVE`. I.e.: + * + * REG (in previous frame) = *(%REG_SAVE + OFFSET) + */ +#define _STR(R) #R +#define __STR(R) _STR(R) +#define __ASM_REG_IS_DEREF_REF_OFFSET(REG_SAVE, REG, OFFSET) \ + ".cfi_escape " __STR(PP_DW_CFA_expression) ", " \ + _STR(REG) "," /* The register being saved */ \ + " 0x02," /* 2 bytes follow */ \ + __STR(PP_DW_OP_breg_base) " + " _STR(REG_SAVE) ", " #OFFSET ";\n\t" + +/* __ASM_REG_IS_DEREF_REF_OFFSET2(REG_SAVE, REG, OFFSET1, OFFSET2) + * + * Like __ASM_REG_IS_DEREF_REF_OFFSET, expect for offsets of more than + * 7 bits. Takes an SLEB128 encoded offset as two bytes. Thus to e.g. + * encode an offset of 0x70, convert to SLEB128 first: + * + * HEX: 0x70 + * BINARY: 01110000 + * SLEB128: 11110000 00000000 + * |||||||| |||||||| + * |||||||| \------- High bits + * |\------ Low bits + * \- Continuation Bit + * + * SLEB128: 0xf0 0x00 + */ +#define __ASM_REG_IS_DEREF_REF_OFFSET2(REG_SAVE, REG, OFFSET1, OFFSET2) \ + ".cfi_escape " __STR(PP_DW_CFA_expression) ", " \ + _STR(REG) "," /* The register being saved */ \ + " 0x03," /* 3 bytes follow */ \ + __STR(PP_DW_OP_breg_base) " + " _STR(REG_SAVE) ", " #OFFSET1 ", " #OFFSET2 ";\n\t" + +/* __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, REG, OFFSET, OFFSET_SUB) + * + * Like __ASM_REG_IS_DEREF_REF_OFFSET, except encodes that the saved + * register value is at `*(%REG_SAVE + OFFSET - OFFSET_SUB). + */ +#define __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, REG, OFFSET, OFFSET_SUB) \ + ".cfi_escape " __STR(PP_DW_CFA_expression) ", " \ + _STR(REG) "," /* The register being saved */ \ + " 0x05," /* 5 bytes follow */ \ + __STR(PP_DW_OP_breg_base) " + " _STR(REG_SAVE) ", " #OFFSET "," \ + __STR(PP_DW_OP_const1u) ", " #OFFSET_SUB "," __STR(PP_DW_OP_minus) ";\n\t" + +/* __ASM_CFA_AT_REG_OFFSET(REG_SAVE, OFFSET) + * + * Declares the CFA of set frame to be stored at *(REG_SAVE + OFFSET). + * Note that the CFA is ordinarily the same as the stack pointer at the end + * of the previous frame, but these are technically separate concepts. + */ +#define __ASM_CFA_AT_REG_OFFSET(REG_SAVE, OFFSET) \ + ".cfi_escape " __STR(PP_DW_CFA_def_cfa_expression) "," \ + " 0x03," /* 3 bytes follow */ \ + __STR(PP_DW_OP_breg_base) " + " _STR(REG_SAVE) ", " #OFFSET ", " \ + __STR(PP_DW_OP_deref) ";\n\t" + +/* __ASM_CFA_AT_REG_OFFSET2(REG_SAVE, OFFSET_HIGH, OFFSET_LOW) + * + * Like __ASM_CFA_AT_REG_OFFSET, except used for larger offsets. `OFFSET_LOW` encodes + * the low 7 bits of the offset. `OFFSET_HIGH` the next 7 bits. + */ +#define __ASM_CFA_AT_REG_OFFSET2(REG_SAVE, OFFSET_HIGH, OFFSET_LOW) \ + ".cfi_escape " __STR(PP_DW_CFA_def_cfa_expression) "," \ + " 0x04," /* 4 bytes follow */ \ + __STR(PP_DW_OP_breg_base) " + " _STR(REG_SAVE) ", " #OFFSET_LOW " | 0x80 , " #OFFSET_HIGH ", " \ + __STR(PP_DW_OP_deref) ";\n\t" + +#define __ASM_REG_IS_DEREF_RCX_OFFSET(REG, OFFSET) \ + __ASM_REG_IS_DEREF_REF_OFFSET(X86_64_DWARF_RCX, REG, OFFSET) +#define __ASM_REG_IS_DEREF_RCX_OFFSET2(REG, OFFSET_HIGH, OFFSET_LOW) \ + __ASM_REG_IS_DEREF_REF_OFFSET2(X86_64_DWARF_RCX, REG, OFFSET_LOW | 0x80, OFFSET_HIGH) + +/* __ASM_REGSAVE_AT_DWARF_REG_MINUS(REG_SAVE, OFFSET_SUB) + * + * Declares that a `struct syscall_frame` containing the previous frame's registers may + * be found at `*(struct syscall_frame*)(%REG_SAVE - OFFSET_SUB)` and declares both the + * CFA and saved registers may be found in the corresponding offset. + */ +#define __ASM_REGSAVE_AT_DWARF_REG_MINUS(REG_SAVE, OFFSET_SUB) \ + /* N.B.: First argument is SLEB128 encoded, so move the high bit over \ + to the subtraction */ \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RIP, 0x00, OFFSET_SUB - 0x70) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RAX, 0x00, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RBX, 0x08, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RDX, 0x18, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RSI, 0x20, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RDI, 0x28, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_R12, 0x00, OFFSET_SUB - 0x50) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_R14, 0x10, OFFSET_SUB - 0x50) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RSP, 0x08, OFFSET_SUB - 0x80) \ + /* -0x10 in 7bit 2's complement */ \ + __ASM_CFA_AT_REG_OFFSET(REG_SAVE, 0x70) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RBP, 0x18, OFFSET_SUB - 0x80) + /*********************************************************************** * __wine_syscall_dispatcher */ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "movq %gs:0x30,%rcx\n\t" + __ASM_CFI(".cfi_undefined rcx\n\t") "movq 0x328(%rcx),%rcx\n\t" /* amd64_thread_data()->syscall_frame */ "popq 0x70(%rcx)\n\t" /* frame->rip */ + __ASM_CFI(".cfi_adjust_cfa_offset -8\n\t") + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_RIP, 0x00, 0x70) "pushfq\n\t" + __ASM_CFI(".cfi_adjust_cfa_offset 8\n\t") "popq 0x80(%rcx)\n\t" + __ASM_CFI(".cfi_adjust_cfa_offset -8\n\t") "movl $0,0x94(%rcx)\n\t" /* frame->restore_flags */ ".globl " __ASM_NAME("__wine_syscall_dispatcher_prolog_end") "\n" __ASM_NAME("__wine_syscall_dispatcher_prolog_end") ":\n\t" "movq %rax,0x00(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RAX, 0x00) "movq %rbx,0x08(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RBX, 0x08) "movq %rdx,0x18(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RDX, 0x18) "movq %rsi,0x20(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RSI, 0x20) "movq %rdi,0x28(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RDI, 0x28) "movq %r12,0x50(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_R12, 0x00, 0x50) "movq %r13,0x58(%rcx)\n\t" "movq %r14,0x60(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_R14, 0x00, 0x60) "movq %r15,0x68(%rcx)\n\t" "movw %cs,0x78(%rcx)\n\t" "movw %ds,0x7a(%rcx)\n\t" "movw %es,0x7c(%rcx)\n\t" "movw %fs,0x7e(%rcx)\n\t" "movq %rsp,0x88(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_RSP, 0x01, 0x08) + __ASM_CFA_AT_REG_OFFSET2(X86_64_DWARF_RCX, 0x01, 0x08) "movw %ss,0x90(%rcx)\n\t" "movw %gs,0x92(%rcx)\n\t" "movq %rbp,0x98(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_RBP, 0x01, 0x18) /* Legends of Runeterra hooks the first system call return instruction, and * depends on us returning to it. Adjust the return address accordingly. */ "subq $0xb,0x70(%rcx)\n\t" @@ -3341,6 +3495,8 @@ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "jmp 3f\n" "2:\tfxsave64 0xc0(%rcx)\n" "3:\tleaq 0x98(%rcx),%rbp\n\t" + __ASM_CFI(".cfi_remember_state\n\t") + __ASM_REGSAVE_AT_DWARF_REG_MINUS(X86_64_DWARF_RBP, 0x98) #ifdef __linux__ "testl $12,%r14d\n\t" /* SYSCALL_HAVE_PTHREAD_TEB | SYSCALL_HAVE_WRFSGSBASE */ "jz 2f\n\t" @@ -3402,35 +3558,60 @@ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "jmp 4f\n" "3:\tfxrstor64 0xc0(%rcx)\n" "4:\tmovq 0x98(%rcx),%rbp\n\t" + __ASM_CFI(".cfi_restore_state\n\t") "movq 0x68(%rcx),%r15\n\t" + __ASM_CFI(".cfi_same_value r15\n\t") "movq 0x60(%rcx),%r14\n\t" + __ASM_CFI(".cfi_same_value r14\n\t") "movq 0x58(%rcx),%r13\n\t" + __ASM_CFI(".cfi_same_value r13\n\t") "movq 0x50(%rcx),%r12\n\t" + __ASM_CFI(".cfi_same_value r12\n\t") "movq 0x28(%rcx),%rdi\n\t" + __ASM_CFI(".cfi_same_value rdi\n\t") "movq 0x20(%rcx),%rsi\n\t" + __ASM_CFI(".cfi_same_value rsi\n\t") "movq 0x08(%rcx),%rbx\n\t" + __ASM_CFI(".cfi_same_value rbx\n\t") "testl $0x3,%edx\n\t" /* CONTEXT_CONTROL | CONTEXT_INTEGER */ "jnz 1f\n\t" + __ASM_CFI(".cfi_remember_state\n\t") "movq 0x80(%rcx),%r11\n\t" /* frame->eflags */ + __ASM_CFI(".cfi_same_value r11\n\t") "pushq %r11\n\t" "popfq\n\t" "movq 0x88(%rcx),%rsp\n\t" + __ASM_CFI(".cfi_same_value rsp\n\t") "movq 0x70(%rcx),%rcx\n\t" /* frame->rip */ + __ASM_CFI(".cfi_register rip, rcx\n\t") "jmpq *%rcx\n\t" "1:\tleaq 0x70(%rcx),%rsp\n\t" + __ASM_CFI(".cfi_restore_state\n\t") + __ASM_CFI(".cfi_remember_state\n\t") "testl $0x2,%edx\n\t" /* CONTEXT_INTEGER */ "jnz 1f\n\t" "movq 0x10(%rsp),%r11\n\t" /* frame->eflags */ + __ASM_CFI(".cfi_same_value r11\n\t") "movq (%rsp),%rcx\n\t" /* frame->rip */ + __ASM_CFI(".cfi_register rip, rcx\n\t") "iretq\n" "1:\tmovq 0x00(%rcx),%rax\n\t" + __ASM_CFI(".cfi_restore_state\n\t") + __ASM_CFI(".cfi_same_value rax\n\t") "movq 0x18(%rcx),%rdx\n\t" + __ASM_CFI(".cfi_same_value rdx\n\t") "movq 0x30(%rcx),%r8\n\t" + __ASM_CFI(".cfi_same_value r8\n\t") "movq 0x38(%rcx),%r9\n\t" + __ASM_CFI(".cfi_same_value r9\n\t") "movq 0x40(%rcx),%r10\n\t" + __ASM_CFI(".cfi_same_value r10\n\t") "movq 0x48(%rcx),%r11\n\t" + __ASM_CFI(".cfi_same_value r11\n\t") "movq 0x10(%rcx),%rcx\n" + __ASM_REG_IS_DEREF_REF_OFFSET(X86_64_DWARF_RSP, X86_64_DWARF_RIP, 0) "iretq\n" + __ASM_REGSAVE_AT_DWARF_REG_MINUS(X86_64_DWARF_RBP, 0x98) "5:\tmovl $0xc000000d,%edx\n\t" /* STATUS_INVALID_PARAMETER */ "movq %rsp,%rcx\n\t" ".globl " __ASM_NAME("__wine_syscall_dispatcher_return") "\n" -- 2.25.1

July 29, 2022

Re: [PATCH] ntdll/signalx86_64: Add CFI for __wine_syscall_dispatcher

by Marvin

Hi, It looks like your patch introduced the new failures shown below. Please investigate and fix them before resubmitting your patch. If they are not new, fixing them anyway would help a lot. Otherwise please ask for the known failures list to be updated. The full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=120103 Your paranoid android. === debian11 (build log) === ../wine/dlls/ntdll/unix/signal_x86_64.c:3339:60: error: '#' is not followed by a macro parameter ../wine/dlls/ntdll/unix/signal_x86_64.c:3361:71: error: '#' is not followed by a macro parameter ../wine/dlls/ntdll/unix/signal_x86_64.c:3372:78: error: '#' is not followed by a macro parameter ../wine/dlls/ntdll/unix/signal_x86_64.c:3385:49: error: '#' is not followed by a macro parameter ../wine/dlls/ntdll/unix/signal_x86_64.c:3396:67: error: '#' is not followed by a macro parameter ../wine/dlls/ntdll/unix/signal_x86_64.c:3405:5: error: expected ‘)’ before ‘__ASM_REG_IS_DEREF_REF_OFFSET2’ Task: The wow64 Wine build failed

July 29, 2022

[PATCH] ntdll/signalx86_64: Add CFI for __wine_syscall_dispatcher

by Keno Fischer

In a debugger that is capable of loading both PE and ELF libraries (e.g. the wine-enhanced GDB version at [1]), this unwind information will allow proper unwind across the __wine_syscall_dispatcher. The CFI itself is relatively straightforward, though the implementation looks a little complex, because support for proper CFI encoding is somewhat limitied in the GNU toolchain. Essentially, as registers are stored in the rcx buffer in this function, we switch their cfi description to point at that rcx buffer, so the registers may be freely used thereafter without affecting the unwind. Eventually the rcx buffer is tracked in rbp instead at which point we pivot the CFI information as well and keep it there for most of the rest of the function. [1] https://github.com/JuliaComputing/gdb-solib-wine Signed-off-by: Keno Fischer <keno(a)juliacomputing.com> --- dlls/ntdll/unix/signal_x86_64.c | 180 ++++++++++++++++++++++++++++++++ 1 file changed, 180 insertions(+) diff --git a/dlls/ntdll/unix/signal_x86_64.c b/dlls/ntdll/unix/signal_x86_64.c index d6f19dbfa84..a6b79335fab 100644 --- a/dlls/ntdll/unix/signal_x86_64.c +++ b/dlls/ntdll/unix/signal_x86_64.c @@ -442,6 +442,10 @@ static USHORT fs32_sel; /* selector for %fs in 32-bit mode */ * Definitions for Dwarf unwind tables */ +/* For use in inline-assembly and other contexts where C enums are unavailable */ +#define PP_DW_CFA_def_cfa_expression 0xf +#define PP_DW_CFA_expression 0x10 + enum dwarf_call_frame_info { DW_CFA_advance_loc = 0x40, @@ -472,6 +476,14 @@ enum dwarf_call_frame_info DW_CFA_val_expression = 0x16, }; +/* DWARF State machine ops */ +#define PP_DW_OP_const1u 0x08 +#define PP_DW_OP_breg_base 0x70 +#define PP_DW_OP_minus 0x1c +#define PP_DW_OP_deref 0x06 +#define PP_DW_OP_plus 0x22 +#define PP_DW_OP_plus_uconst 0x23 + enum dwarf_operation { DW_OP_addr = 0x03, @@ -3288,35 +3300,176 @@ __ASM_GLOBAL_FUNC( signal_exit_thread, __ASM_CFI(".cfi_rel_offset %r15,8\n\t") "call *%rsi" ) + +/*********************************************************************** + * CFI Utility definitions + * + * These utility definitions help encode CFI unwind information (for consumption + * in gdb and similar debuggers), improving the accuracy of backtraces. + */ + +/* DWARF Register ordering for x86_64 */ +#define X86_64_DWARF_RAX 0x00 +#define X86_64_DWARF_RDX 0x01 +#define X86_64_DWARF_RCX 0x02 +#define X86_64_DWARF_RBX 0x03 +#define X86_64_DWARF_RSI 0x04 +#define X86_64_DWARF_RDI 0x05 +#define X86_64_DWARF_RBP 0x06 +#define X86_64_DWARF_RSP 0x07 +#define X86_64_DWARF_R8 0x08 +#define X86_64_DWARF_R9 0x09 +#define X86_64_DWARF_R10 0x0a +#define X86_64_DWARF_R11 0x0b +#define X86_64_DWARF_R12 0x0c +#define X86_64_DWARF_R13 0x0d +#define X86_64_DWARF_R14 0x0e +#define X86_64_DWARF_R15 0x0f +#define X86_64_DWARF_RIP 0x10 + +/* __ASM_REG_IS_DEREF_REF_OFFSET(REG_SAVE, REG, OFFSET) + * + * Declare that the register with DWARF register number `REG` is + * saved at offset `OFFSET` from the register with DWARF register + * number `REG_SAVE`. I.e.: + * + * REG (in previous frame) = *(%REG_SAVE + OFFSET) + */ +#define _STR(R) #R +#define __ASM_REG_IS_DEREF_REF_OFFSET(REG_SAVE, REG, OFFSET) \ + ".cfi_escape " #PP_DW_CFA_expression", " \ + _STR(REG) "," /* The register being saved */ \ + " 0x02," /* 2 bytes follow */ \ + #PP_DW_OP_breg_base " + " _STR(REG_SAVE) ", " #OFFSET ";\n\t" + +/* __ASM_REG_IS_DEREF_REF_OFFSET2(REG_SAVE, REG, OFFSET1, OFFSET2) + * + * Like __ASM_REG_IS_DEREF_REF_OFFSET, expect for offsets of more than + * 7 bits. Takes an SLEB128 encoded offset as two bytes. Thus to e.g. + * encode an offset of 0x70, convert to SLEB128 first: + * + * HEX: 0x70 + * BINARY: 01110000 + * SLEB128: 11110000 00000000 + * |||||||| |||||||| + * |||||||| \------- High bits + * |\------ Low bits + * \- Continuation Bit + * + * SLEB128: 0xf0 0x00 + */ +#define __ASM_REG_IS_DEREF_REF_OFFSET2(REG_SAVE, REG, OFFSET1, OFFSET2) \ + ".cfi_escape " #PP_DW_CFA_expression", " \ + _STR(REG) "," /* The register being saved */ \ + " 0x03," /* 3 bytes follow */ \ + #PP_DW_OP_breg_base " + " _STR(REG_SAVE) ", " #OFFSET1 ", " #OFFSET2 ";\n\t" + +/* __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, REG, OFFSET, OFFSET_SUB) + * + * Like __ASM_REG_IS_DEREF_REF_OFFSET, except encodes that the saved + * register value is at `*(%REG_SAVE + OFFSET - OFFSET_SUB). + */ +#define __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, REG, OFFSET, OFFSET_SUB) \ + ".cfi_escape " #PP_DW_CFA_expression", " \ + _STR(REG) "," /* The register being saved */ \ + " 0x05," /* 5 bytes follow */ \ + #PP_DW_OP_breg_base " + " _STR(REG_SAVE) ", " #OFFSET "," \ + #PP_DW_OP_const1u ", " #OFFSET_SUB "," #PP_DW_OP_minus ";\n\t" + +/* __ASM_CFA_AT_REG_OFFSET(REG_SAVE, OFFSET) + * + * Declares the CFA of set frame to be stored at *(REG_SAVE + OFFSET). + * Note that the CFA is ordinarily the same as the stack pointer at the end + * of the previous frame, but these are technically separate concepts. + */ +#define __ASM_CFA_AT_REG_OFFSET(REG_SAVE, OFFSET) \ + ".cfi_escape " #PP_DW_CFA_def_cfa_expression "," \ + " 0x03," /* 3 bytes follow */ \ + #PP_DW_OP_breg_base " + " _STR(REG_SAVE) ", " #OFFSET ", " \ + #PP_DW_OP_deref ";\n\t" + +/* __ASM_CFA_AT_REG_OFFSET2(REG_SAVE, OFFSET_HIGH, OFFSET_LOW) + * + * Like __ASM_CFA_AT_REG_OFFSET, except used for larger offsets. `OFFSET_LOW` encodes + * the low 7 bits of the offset. `OFFSET_HIGH` the next 7 bits. + */ +#define __ASM_CFA_AT_REG_OFFSET2(REG_SAVE, OFFSET_HIGH, OFFSET_LOW) \ + ".cfi_escape " DW_CFA_def_cfa_expression "," \ + " 0x04," /* 4 bytes follow */ \ + #PP_DW_OP_breg_base " + " _STR(REG_SAVE) ", " #OFFSET_LOW " | 0x80 , " #OFFSET_HIGH ", " \ + #PP_DW_OP_deref ";\n\t" + +#define __ASM_REG_IS_DEREF_RCX_OFFSET(REG, OFFSET) \ + __ASM_REG_IS_DEREF_REF_OFFSET(X86_64_DWARF_RCX, REG, OFFSET) +#define __ASM_REG_IS_DEREF_RCX_OFFSET2(REG, OFFSET_HIGH, OFFSET_LOW) \ + __ASM_REG_IS_DEREF_REF_OFFSET2(X86_64_DWARF_RCX, REG, OFFSET_LOW | 0x80, OFFSET_HIGH) + +/* __ASM_REGSAVE_AT_DWARF_REG_MINUS(REG_SAVE, OFFSET_SUB) + * + * Declares that a `struct syscall_frame` containing the previous frame's registers may + * be found at `*(struct syscall_frame*)(%REG_SAVE - OFFSET_SUB)` and declares both the + * CFA and saved registers may be found in the corresponding offset. + */ +#define __ASM_REGSAVE_AT_DWARF_REG_MINUS(REG_SAVE, OFFSET_SUB) \ + /* N.B.: First argument is SLEB128 encoded, so move the high bit over \ + to the subtraction */ \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RIP, 0x00, OFFSET_SUB - 0x70) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RAX, 0x00, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RBX, 0x08, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RDX, 0x18, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RSI, 0x20, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RDI, 0x28, OFFSET_SUB) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_R12, 0x00, OFFSET_SUB - 0x50) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_R14, 0x10, OFFSET_SUB - 0x50) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RSP, 0x08, OFFSET_SUB - 0x80) \ + /* -0x10 in 7bit 2's complement */ \ + __ASM_CFA_AT_REG_OFFSET(REG_SAVE, 0x70) \ + __ASM_REG_IS_DEREF_REF_OFFSET_MINUS(REG_SAVE, X86_64_DWARF_RBP, 0x18, OFFSET_SUB - 0x80) + /*********************************************************************** * __wine_syscall_dispatcher */ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "movq %gs:0x30,%rcx\n\t" + __ASM_CFI(".cfi_undefined rcx\n\t") "movq 0x328(%rcx),%rcx\n\t" /* amd64_thread_data()->syscall_frame */ "popq 0x70(%rcx)\n\t" /* frame->rip */ + __ASM_CFI(".cfi_adjust_cfa_offset -8\n\t") + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_RIP, 0x00, 0x70) "pushfq\n\t" + __ASM_CFI(".cfi_adjust_cfa_offset 8\n\t") "popq 0x80(%rcx)\n\t" + __ASM_CFI(".cfi_adjust_cfa_offset -8\n\t") "movl $0,0x94(%rcx)\n\t" /* frame->restore_flags */ ".globl " __ASM_NAME("__wine_syscall_dispatcher_prolog_end") "\n" __ASM_NAME("__wine_syscall_dispatcher_prolog_end") ":\n\t" "movq %rax,0x00(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RAX, 0x00) "movq %rbx,0x08(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RBX, 0x08) "movq %rdx,0x18(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RDX, 0x18) "movq %rsi,0x20(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RSI, 0x20) "movq %rdi,0x28(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET(X86_64_DWARF_RDI, 0x28) "movq %r12,0x50(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_R12, 0x00, 0x50) "movq %r13,0x58(%rcx)\n\t" "movq %r14,0x60(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_R14, 0x00, 0x60) "movq %r15,0x68(%rcx)\n\t" "movw %cs,0x78(%rcx)\n\t" "movw %ds,0x7a(%rcx)\n\t" "movw %es,0x7c(%rcx)\n\t" "movw %fs,0x7e(%rcx)\n\t" "movq %rsp,0x88(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_RSP, 0x01, 0x08) + __ASM_CFA_AT_REG_OFFSET2(X86_64_DWARF_RCX, 0x01, 0x08) "movw %ss,0x90(%rcx)\n\t" "movw %gs,0x92(%rcx)\n\t" "movq %rbp,0x98(%rcx)\n\t" + __ASM_REG_IS_DEREF_RCX_OFFSET2(X86_64_DWARF_RBP, 0x01, 0x18) /* Legends of Runeterra hooks the first system call return instruction, and * depends on us returning to it. Adjust the return address accordingly. */ "subq $0xb,0x70(%rcx)\n\t" @@ -3341,6 +3494,8 @@ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "jmp 3f\n" "2:\tfxsave64 0xc0(%rcx)\n" "3:\tleaq 0x98(%rcx),%rbp\n\t" + __ASM_CFI(".cfi_remember_state\n\t") + __ASM_REGSAVE_AT_DWARF_REG_MINUS(X86_64_DWARF_RBP, 0x98) #ifdef __linux__ "testl $12,%r14d\n\t" /* SYSCALL_HAVE_PTHREAD_TEB | SYSCALL_HAVE_WRFSGSBASE */ "jz 2f\n\t" @@ -3402,35 +3557,60 @@ __ASM_GLOBAL_FUNC( __wine_syscall_dispatcher, "jmp 4f\n" "3:\tfxrstor64 0xc0(%rcx)\n" "4:\tmovq 0x98(%rcx),%rbp\n\t" + __ASM_CFI(".cfi_restore_state\n\t") "movq 0x68(%rcx),%r15\n\t" + __ASM_CFI(".cfi_same_value r15\n\t") "movq 0x60(%rcx),%r14\n\t" + __ASM_CFI(".cfi_same_value r14\n\t") "movq 0x58(%rcx),%r13\n\t" + __ASM_CFI(".cfi_same_value r13\n\t") "movq 0x50(%rcx),%r12\n\t" + __ASM_CFI(".cfi_same_value r12\n\t") "movq 0x28(%rcx),%rdi\n\t" + __ASM_CFI(".cfi_same_value rdi\n\t") "movq 0x20(%rcx),%rsi\n\t" + __ASM_CFI(".cfi_same_value rsi\n\t") "movq 0x08(%rcx),%rbx\n\t" + __ASM_CFI(".cfi_same_value rbx\n\t") "testl $0x3,%edx\n\t" /* CONTEXT_CONTROL | CONTEXT_INTEGER */ "jnz 1f\n\t" + __ASM_CFI(".cfi_remember_state\n\t") "movq 0x80(%rcx),%r11\n\t" /* frame->eflags */ + __ASM_CFI(".cfi_same_value r11\n\t") "pushq %r11\n\t" "popfq\n\t" "movq 0x88(%rcx),%rsp\n\t" + __ASM_CFI(".cfi_same_value rsp\n\t") "movq 0x70(%rcx),%rcx\n\t" /* frame->rip */ + __ASM_CFI(".cfi_register rip, rcx\n\t") "jmpq *%rcx\n\t" "1:\tleaq 0x70(%rcx),%rsp\n\t" + __ASM_CFI(".cfi_restore_state\n\t") + __ASM_CFI(".cfi_remember_state\n\t") "testl $0x2,%edx\n\t" /* CONTEXT_INTEGER */ "jnz 1f\n\t" "movq 0x10(%rsp),%r11\n\t" /* frame->eflags */ + __ASM_CFI(".cfi_same_value r11\n\t") "movq (%rsp),%rcx\n\t" /* frame->rip */ + __ASM_CFI(".cfi_register rip, rcx\n\t") "iretq\n" "1:\tmovq 0x00(%rcx),%rax\n\t" + __ASM_CFI(".cfi_restore_state\n\t") + __ASM_CFI(".cfi_same_value rax\n\t") "movq 0x18(%rcx),%rdx\n\t" + __ASM_CFI(".cfi_same_value rdx\n\t") "movq 0x30(%rcx),%r8\n\t" + __ASM_CFI(".cfi_same_value r8\n\t") "movq 0x38(%rcx),%r9\n\t" + __ASM_CFI(".cfi_same_value r9\n\t") "movq 0x40(%rcx),%r10\n\t" + __ASM_CFI(".cfi_same_value r10\n\t") "movq 0x48(%rcx),%r11\n\t" + __ASM_CFI(".cfi_same_value r11\n\t") "movq 0x10(%rcx),%rcx\n" + __ASM_REG_IS_DEREF_REF_OFFSET(X86_64_DWARF_RSP, X86_64_DWARF_RIP, 0) "iretq\n" + __ASM_REGSAVE_AT_DWARF_REG_MINUS(X86_64_DWARF_RBP, 0x98) "5:\tmovl $0xc000000d,%edx\n\t" /* STATUS_INVALID_PARAMETER */ "movq %rsp,%rcx\n\t" ".globl " __ASM_NAME("__wine_syscall_dispatcher_return") "\n" -- 2.25.1

July 29, 2022

Re: [PATCH v4 01/11] winmm/tests: Add tests for visibility of video window.

by Marvin

Hi, It looks like your patch introduced the new failures shown below. Please investigate and fix them before resubmitting your patch. If they are not new, fixing them anyway would help a lot. Otherwise please ask for the known failures list to be updated. The full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=120073 Your paranoid android. === w7u_adm (32 bit report) === winmm: midi.c:1039: Test failed: bad message 1e/0 from midiStreamOut, expect 3c9/22f940 midi.c:1048: Test failed: bad message 3c9/22f940 from midiStreamClose, expect 3c8/0 === w7u_el (32 bit report) === winmm: midi.c:988: Test failed: bad message 1e/0 from midiStreamOpen, expect 3c7/0 midi.c:1039: Test failed: bad message 3c7/0 from midiStreamOut, expect 3c9/22f940 midi.c:1048: Test failed: bad message 3c9/22f940 from midiStreamClose, expect 3c8/0

July 28, 2022

Re: [PATCH v3 2/2] ddraw/tests: Add some video memory accounting tests.

by Marvin

Hi, It looks like your patch introduced the new failures shown below. Please investigate and fix them before resubmitting your patch. If they are not new, fixing them anyway would help a lot. Otherwise please ask for the known failures list to be updated. The tests also ran into some preexisting test failures. If you know how to fix them that would be helpful. See the TestBot job for the details: The full results can be found at: https://testbot.winehq.org/JobDetails.pl?Key=120071 Your paranoid android. === w8adm (32 bit report) === ddraw: ddraw7.c:18888: Test failed: Got unexpected color 0x0000ff00. === w864 (32 bit report) === ddraw: ddraw7.c:18888: Test failed: Got unexpected color 0x0000ff00. === w1064_tsign (32 bit report) === ddraw: ddraw7.c:18888: Test failed: Got unexpected color 0x00000040. === debian11 (32 bit Chinese:China report) === Report validation errors: ddraw1: Timeout === debian11 (build log) === WineRunWineTest.pl:error: The task timed out

July 28, 2022

Re: [PATCH vkd3d v3 17/17] vkd3d-shader/hlsl: Replace register offsets with index paths in split copies.

by Matteo Bruni

On Wed, Jul 20, 2022 at 3:24 PM Francisco Casas <fcasas(a)codeweavers.com> wrote: > > hlsl_new_store() and hlsl_new_load() are deleted, so now there are no more > direct ways to create derefs with offsets in hlsl.c and hlsl.h. > > Signed-off-by: Francisco Casas <fcasas(a)codeweavers.com> > --- > > v3: > * No changes. > > Signed-off-by: Francisco Casas <fcasas(a)codeweavers.com> > --- > libs/vkd3d-shader/hlsl.c | 32 ------------------------- > libs/vkd3d-shader/hlsl.h | 4 ---- > libs/vkd3d-shader/hlsl_codegen.c | 40 +++++++++----------------------- > 3 files changed, 11 insertions(+), 65 deletions(-) > > diff --git a/libs/vkd3d-shader/hlsl.c b/libs/vkd3d-shader/hlsl.c > index 20dcd9af..6b187e7d 100644 > --- a/libs/vkd3d-shader/hlsl.c > +++ b/libs/vkd3d-shader/hlsl.c > @@ -795,25 +795,6 @@ void hlsl_cleanup_deref(struct hlsl_deref *deref) > hlsl_src_remove(&deref->offset); > } > > -struct hlsl_ir_store *hlsl_new_store(struct hlsl_ctx *ctx, struct hlsl_ir_var *var, struct hlsl_ir_node *offset, > - struct hlsl_ir_node *rhs, unsigned int writemask, struct vkd3d_shader_location loc) > -{ > - struct hlsl_ir_store *store; > - > - if (!writemask && type_is_single_reg(rhs->data_type)) > - writemask = (1 << rhs->data_type->dimx) - 1; > - > - if (!(store = hlsl_alloc(ctx, sizeof(*store)))) > - return NULL; > - > - init_node(&store->node, HLSL_IR_STORE, NULL, loc); > - init_deref(ctx, &store->lhs, var, 0); > - hlsl_src_from_node(&store->lhs.offset, offset); > - hlsl_src_from_node(&store->rhs, rhs); > - store->writemask = writemask; > - return store; > -} > - > /* Initializes a simple variable derefence, so that it can be passed to load/store functions. */ > void hlsl_init_simple_deref_from_var(struct hlsl_deref *deref, struct hlsl_ir_var *var) > { > @@ -969,19 +950,6 @@ struct hlsl_ir_if *hlsl_new_if(struct hlsl_ctx *ctx, struct hlsl_ir_node *condit > return iff; > } > > -struct hlsl_ir_load *hlsl_new_load(struct hlsl_ctx *ctx, struct hlsl_ir_var *var, struct hlsl_ir_node *offset, > - struct hlsl_type *type, const struct vkd3d_shader_location loc) > -{ > - struct hlsl_ir_load *load; > - > - if (!(load = hlsl_alloc(ctx, sizeof(*load)))) > - return NULL; > - init_node(&load->node, HLSL_IR_LOAD, type, loc); > - init_deref(ctx, &load->src, var, 0); > - hlsl_src_from_node(&load->src.offset, offset); > - return load; > -} > - > struct hlsl_ir_load *hlsl_new_load_index(struct hlsl_ctx *ctx, const struct hlsl_deref *deref, > struct hlsl_ir_node *idx, const struct vkd3d_shader_location *loc) > { > diff --git a/libs/vkd3d-shader/hlsl.h b/libs/vkd3d-shader/hlsl.h > index e1ddd0cf..77005407 100644 > --- a/libs/vkd3d-shader/hlsl.h > +++ b/libs/vkd3d-shader/hlsl.h > @@ -774,11 +774,7 @@ struct hlsl_ir_resource_load *hlsl_new_resource_load(struct hlsl_ctx *ctx, struc > enum hlsl_resource_load_type type, struct hlsl_deref *resource, struct hlsl_deref *sampler, > struct hlsl_ir_node *coords, struct hlsl_ir_node *texel_offset, const struct vkd3d_shader_location *loc); > > -struct hlsl_ir_load *hlsl_new_load(struct hlsl_ctx *ctx, struct hlsl_ir_var *var, struct hlsl_ir_node *offset, > - struct hlsl_type *type, struct vkd3d_shader_location loc); > struct hlsl_ir_loop *hlsl_new_loop(struct hlsl_ctx *ctx, struct vkd3d_shader_location loc); > -struct hlsl_ir_store *hlsl_new_store(struct hlsl_ctx *ctx, struct hlsl_ir_var *var, struct hlsl_ir_node *offset, > - struct hlsl_ir_node *rhs, unsigned int writemask, struct vkd3d_shader_location loc); > struct hlsl_type *hlsl_new_struct_type(struct hlsl_ctx *ctx, const char *name, > struct hlsl_struct_field *fields, size_t field_count); > struct hlsl_ir_swizzle *hlsl_new_swizzle(struct hlsl_ctx *ctx, DWORD s, unsigned int components, > diff --git a/libs/vkd3d-shader/hlsl_codegen.c b/libs/vkd3d-shader/hlsl_codegen.c > index 7a245007..08029361 100644 > --- a/libs/vkd3d-shader/hlsl_codegen.c > +++ b/libs/vkd3d-shader/hlsl_codegen.c > @@ -950,39 +950,21 @@ static bool fold_redundant_casts(struct hlsl_ctx *ctx, struct hlsl_ir_node *inst > * split_matrix_copies(). Inserts new instructions right before > * "store". */ > static bool split_copy(struct hlsl_ctx *ctx, struct hlsl_ir_store *store, > - const struct hlsl_ir_load *load, const unsigned int offset, struct hlsl_type *type) > + const struct hlsl_ir_load *load, const unsigned int indx, struct hlsl_type *type) Why "indx" now?

July 28, 2022