Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
From: Peter Maydell [peter.mayd...@linaro.org]
Sent: Thursday, June 27, 2013 7:21 PM
To: Petar Jovanovic
Cc: qemu-devel@nongnu.org; blauwir...@gmail.com; riku.voi...@linaro.org; Petar
Jovanovic; aurel...@aurel32.net; r...@twiddle.net
Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve
target_to_host_sock_type conversion
On 1 April 2013 16:49, Petar Jovanovic petar.jovano...@rt-rk.com wrote:
From: Petar Jovanovic petar.jovano...@imgtec.com
Previous implementation has failed to take into account different value of
SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC.
The same conversion has to be applied both for do_socket and do_socketpair,
so the code has been isolated in a static inline function.
The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES,
these are MIPS, SPARC and ALPHA now.
+#if defined(ARCH_HAS_SOCKET_TYPES)
+static inline void target_to_host_sock_type(int *type)
{
[etc]
I apologise for the hugely late review on this patch, but
this looks odd. Whether we need to translate SOCK_* constants
doesn't just depend on the target architecture, but also on
the host. (Consider running QEMU on MIPS and emulating x86.)
I think that we need to make sure that we define TARGET_SOCK_*
for all architectures (ie with a bit in the common section
that goes #ifndef ARCH_HAS_SOCKET_TYPES [default stuff]),
and then unconditionally define and call target_to_host_sock_type
in syscall.c.
thanks
-- PMM
I have just resubmitted the patch with this addition included.
Check for [PATCH v3].
Thanks.
Petar
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
On 1 April 2013 16:49, Petar Jovanovic petar.jovano...@rt-rk.com wrote:
From: Petar Jovanovic petar.jovano...@imgtec.com
Previous implementation has failed to take into account different value of
SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC.
The same conversion has to be applied both for do_socket and do_socketpair,
so the code has been isolated in a static inline function.
The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES,
these are MIPS, SPARC and ALPHA now.
+#if defined(ARCH_HAS_SOCKET_TYPES)
+static inline void target_to_host_sock_type(int *type)
{
[etc]
I apologise for the hugely late review on this patch, but
this looks odd. Whether we need to translate SOCK_* constants
doesn't just depend on the target architecture, but also on
the host. (Consider running QEMU on MIPS and emulating x86.)
I think that we need to make sure that we define TARGET_SOCK_*
for all architectures (ie with a bit in the common section
that goes #ifndef ARCH_HAS_SOCKET_TYPES [default stuff]),
and then unconditionally define and call target_to_host_sock_type
in syscall.c.
thanks
-- PMM
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
Hi, On 27 May 2013 13:49, Petar Jovanovic petar.jovano...@imgtec.com wrote: Can anyone take a look at this and commit it if there are no other change requests? I thought Aurelian had an issue with this patch, but it seems you explained your position well. I'll get this included in my next pull. Riku From: Petar Jovanovic Sent: Wednesday, May 08, 2013 1:16 AM To: riku.voi...@linaro.org; qemu-devel@nongnu.org Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; r...@twiddle.net; Alexander Graf; Andreas Färber Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic Sent: Tuesday, April 30, 2013 3:20 AM To: Andreas Färber Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion My assumption was that if new socket values are added in future, they will likely be the same for all platforms, so they can be supported without adding new lines of code. Here we convert the values known to be different, we leave the values known to be the same, and for the incorrect values - we pass them to (host) kernel as they are in belief that kernel would return error. We could handle that part and check for incorrect value before passing it to the kernel, but in that case we bring more kernel logic to QEMU for the cases that are not likely to happen. Petar From: Andreas Färber [afaer...@suse.de] Sent: Monday, April 29, 2013 4:56 PM To: Petar Jovanovic Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
Thanks. Petar From: Riku Voipio [riku.voi...@linaro.org] Sent: Thursday, May 30, 2013 3:28 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; r...@twiddle.net; Alexander Graf; Andreas Färber Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Hi, On 27 May 2013 13:49, Petar Jovanovic petar.jovano...@imgtec.com wrote: Can anyone take a look at this and commit it if there are no other change requests? I thought Aurelian had an issue with this patch, but it seems you explained your position well. I'll get this included in my next pull. Riku From: Petar Jovanovic Sent: Wednesday, May 08, 2013 1:16 AM To: riku.voi...@linaro.org; qemu-devel@nongnu.org Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; r...@twiddle.net; Alexander Graf; Andreas Färber Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic Sent: Tuesday, April 30, 2013 3:20 AM To: Andreas Färber Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion My assumption was that if new socket values are added in future, they will likely be the same for all platforms, so they can be supported without adding new lines of code. Here we convert the values known to be different, we leave the values known to be the same, and for the incorrect values - we pass them to (host) kernel as they are in belief that kernel would return error. We could handle that part and check for incorrect value before passing it to the kernel, but in that case we bring more kernel logic to QEMU for the cases that are not likely to happen. Petar From: Andreas Färber [afaer...@suse.de] Sent: Monday, April 29, 2013 4:56 PM To: Petar Jovanovic Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
Can anyone take a look at this and commit it if there are no other change requests? Thank you. From: Petar Jovanovic Sent: Wednesday, May 08, 2013 1:16 AM To: riku.voi...@linaro.org; qemu-devel@nongnu.org Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; r...@twiddle.net; Alexander Graf; Andreas Färber Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic Sent: Tuesday, April 30, 2013 3:20 AM To: Andreas Färber Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion My assumption was that if new socket values are added in future, they will likely be the same for all platforms, so they can be supported without adding new lines of code. Here we convert the values known to be different, we leave the values known to be the same, and for the incorrect values - we pass them to (host) kernel as they are in belief that kernel would return error. We could handle that part and check for incorrect value before passing it to the kernel, but in that case we bring more kernel logic to QEMU for the cases that are not likely to happen. Petar From: Andreas Färber [afaer...@suse.de] Sent: Monday, April 29, 2013 4:56 PM To: Petar Jovanovic Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
ping From: Petar Jovanovic Sent: Wednesday, May 08, 2013 1:16 AM To: riku.voi...@linaro.org; qemu-devel@nongnu.org Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; r...@twiddle.net; Alexander Graf; Andreas Färber Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic Sent: Tuesday, April 30, 2013 3:20 AM To: Andreas Färber Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion My assumption was that if new socket values are added in future, they will likely be the same for all platforms, so they can be supported without adding new lines of code. Here we convert the values known to be different, we leave the values known to be the same, and for the incorrect values - we pass them to (host) kernel as they are in belief that kernel would return error. We could handle that part and check for incorrect value before passing it to the kernel, but in that case we bring more kernel logic to QEMU for the cases that are not likely to happen. Petar From: Andreas Färber [afaer...@suse.de] Sent: Monday, April 29, 2013 4:56 PM To: Petar Jovanovic Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic Sent: Tuesday, April 30, 2013 3:20 AM To: Andreas Färber Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: RE: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion My assumption was that if new socket values are added in future, they will likely be the same for all platforms, so they can be supported without adding new lines of code. Here we convert the values known to be different, we leave the values known to be the same, and for the incorrect values - we pass them to (host) kernel as they are in belief that kernel would return error. We could handle that part and check for incorrect value before passing it to the kernel, but in that case we bring more kernel logic to QEMU for the cases that are not likely to happen. Petar From: Andreas Färber [afaer...@suse.de] Sent: Monday, April 29, 2013 4:56 PM To: Petar Jovanovic Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic Sent: Tuesday, April 23, 2013 2:15 AM To: Aurelien Jarno; Petar Jovanovic Cc: qemu-devel@nongnu.org; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: RE: [PATCH v2] linux-user: improve target_to_host_sock_type conversion Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? Petar From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ - #define TARGET_SO_STYLESO_TYPE /* Synonym */ - #define TARGET_SO_ERROR0x1007 /* get error status and clear */ - #define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ - #define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ - #define TARGET_SO_SNDLOWAT 0x1003 /* send low-water mark */ - #define TARGET_SO_RCVLOWAT 0x1004 /* receive low-water mark */ - #define TARGET_SO_SNDTIMEO 0x1005 /* send timeout */ - #define TARGET_SO_RCVTIMEO 0x1006 /* receive timeout */ - #define TARGET_SO_ACCEPTCONN 0x1009 - - /* linux-specific, might as well be the same as on i386 */ - #define TARGET_SO_NO_CHECK 11 - #define TARGET_SO_PRIORITY 12 - #define TARGET_SO_BSDCOMPAT14 +/* MIPS special values for constants */ + +/* + * For setsockopt(2) + * + * This defines are ABI conformant as far as Linux supports these ... + */ +#define TARGET_SOL_SOCKET 0x + +#define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ +#define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ +#define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send + SIGPIPE when they die. */ +#define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ +#define TARGET_SO_BROADCAST0x0020 /* Allow transmission of + broadcast messages. */ +#define TARGET_SO_LINGER
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ - #define TARGET_SO_STYLESO_TYPE /* Synonym */ - #define TARGET_SO_ERROR0x1007 /* get error status and clear */ - #define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ - #define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ - #define TARGET_SO_SNDLOWAT 0x1003 /* send low-water mark */ - #define TARGET_SO_RCVLOWAT 0x1004 /* receive low-water mark */ - #define TARGET_SO_SNDTIMEO 0x1005 /* send timeout */ - #define TARGET_SO_RCVTIMEO 0x1006 /* receive timeout */ - #define TARGET_SO_ACCEPTCONN 0x1009 - - /* linux-specific, might as well be the same as on i386 */ - #define TARGET_SO_NO_CHECK 11 - #define TARGET_SO_PRIORITY 12 - #define TARGET_SO_BSDCOMPAT14 +/* MIPS special values for constants */ + +/* + * For setsockopt(2) + * + * This defines are ABI conformant as far as Linux supports these ... + */ +#define TARGET_SOL_SOCKET 0x + +#define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ +#define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ +#define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send + SIGPIPE when they die. */ +
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
My assumption was that if new socket values are added in future, they will likely be the same for all platforms, so they can be supported without adding new lines of code. Here we convert the values known to be different, we leave the values known to be the same, and for the incorrect values - we pass them to (host) kernel as they are in belief that kernel would return error. We could handle that part and check for incorrect value before passing it to the kernel, but in that case we bring more kernel logic to QEMU for the cases that are not likely to happen. Petar From: Andreas Färber [afaer...@suse.de] Sent: Monday, April 29, 2013 4:56 PM To: Petar Jovanovic Cc: Aurelien Jarno; Petar Jovanovic; blauwir...@gmail.com; riku.voi...@linaro.org; qemu-devel@nongnu.org; r...@twiddle.net; Alexander Graf Subject: Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion Am 23.04.2013 02:15, schrieb Petar Jovanovic: Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? No objection but isn't that a frequent issue that mappings may need to be extended from time to time? The way I've seen that handled is on a case by case basis mapping from one known value to another, with defaulting to whatever form of error reporting appropriate. Here it seems that some cases were dropped and we are now defaulting to taking the literal value where no difference is known. This may lead to silent errors, whereas an abort as other extreme may prohibit use cases with no value difference between host and target. Andreas From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ - #define TARGET_SO_STYLESO_TYPE /* Synonym */ - #define TARGET_SO_ERROR0x1007 /* get error status and clear */ - #define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ - #define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ - #define TARGET_SO_SNDLOWAT 0x1003 /* send low-water mark */ - #define TARGET_SO_RCVLOWAT 0x1004
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
Thanks Aurelien for your comments. What others think? Can we submit this version of the patch? Riku? Richard, Blue? Petar From: Aurelien Jarno [aurel...@aurel32.net] Sent: Monday, April 15, 2013 3:47 PM To: Petar Jovanovic Cc: qemu-devel@nongnu.org; Petar Jovanovic; riku.voi...@linaro.org; r...@twiddle.net; blauwir...@gmail.com Subject: Re: [PATCH v2] linux-user: improve target_to_host_sock_type conversion On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ - #define TARGET_SO_STYLESO_TYPE /* Synonym */ - #define TARGET_SO_ERROR0x1007 /* get error status and clear */ - #define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ - #define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ - #define TARGET_SO_SNDLOWAT 0x1003 /* send low-water mark */ - #define TARGET_SO_RCVLOWAT 0x1004 /* receive low-water mark */ - #define TARGET_SO_SNDTIMEO 0x1005 /* send timeout */ - #define TARGET_SO_RCVTIMEO 0x1006 /* receive timeout */ - #define TARGET_SO_ACCEPTCONN 0x1009 - - /* linux-specific, might as well be the same as on i386 */ - #define TARGET_SO_NO_CHECK 11 - #define TARGET_SO_PRIORITY 12 - #define TARGET_SO_BSDCOMPAT14 +/* MIPS special values for constants */ + +/* + * For setsockopt(2) + * + * This defines are ABI conformant as far as Linux supports these ... + */ +#define TARGET_SOL_SOCKET 0x + +#define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ +#define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ +#define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send + SIGPIPE when they die. */ +#define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ +#define TARGET_SO_BROADCAST0x0020 /* Allow transmission of + broadcast messages. */ +#define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable +* socket to transmit pending data. +*/ +#define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. +*/ +#if 0 +/* To add: Allow local
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
On Mon, Apr 01, 2013 at 05:49:39PM +0200, Petar Jovanovic wrote: From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* - * For setsockopt(2) - * - * This defines are ABI conformant as far as Linux supports these ... - */ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ - #define TARGET_SO_STYLESO_TYPE /* Synonym */ - #define TARGET_SO_ERROR0x1007 /* get error status and clear */ - #define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ - #define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ - #define TARGET_SO_SNDLOWAT 0x1003 /* send low-water mark */ - #define TARGET_SO_RCVLOWAT 0x1004 /* receive low-water mark */ - #define TARGET_SO_SNDTIMEO 0x1005 /* send timeout */ - #define TARGET_SO_RCVTIMEO 0x1006 /* receive timeout */ - #define TARGET_SO_ACCEPTCONN 0x1009 - - /* linux-specific, might as well be the same as on i386 */ - #define TARGET_SO_NO_CHECK 11 - #define TARGET_SO_PRIORITY 12 - #define TARGET_SO_BSDCOMPAT14 +/* MIPS special values for constants */ + +/* + * For setsockopt(2) + * + * This defines are ABI conformant as far as Linux supports these ... + */ +#define TARGET_SOL_SOCKET 0x + +#define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ +#define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ +#define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send + SIGPIPE when they die. */ +#define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ +#define TARGET_SO_BROADCAST0x0020 /* Allow transmission of + broadcast messages. */ +#define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable +* socket to transmit pending data. +*/ +#define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. +*/ +#if 0 +/* To add: Allow local address and port reuse. */ +#define TARGET_SO_REUSEPORT 0x0200 +#endif + +#define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ +#define TARGET_SO_STYLESO_TYPE /* Synonym */ +#define TARGET_SO_ERROR0x1007 /* get error status and clear */ +#define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ +#define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ +
Re: [Qemu-devel] [PATCH v2] linux-user: improve target_to_host_sock_type conversion
ping http://patchwork.ozlabs.org/patch/232770/ From: Petar Jovanovic [petar.jovano...@rt-rk.com] Sent: Monday, April 01, 2013 5:49 PM To: qemu-devel@nongnu.org Cc: Petar Jovanovic; riku.voi...@linaro.org; aurel...@aurel32.net; r...@twiddle.net; blauwir...@gmail.com Subject: [PATCH v2] linux-user: improve target_to_host_sock_type conversion From: Petar Jovanovic petar.jovano...@imgtec.com Previous implementation has failed to take into account different value of SOCK_NONBLOCK on target and host, and existence of SOCK_CLOEXEC. The same conversion has to be applied both for do_socket and do_socketpair, so the code has been isolated in a static inline function. The change is valid for architectures that define ARCH_HAS_SOCKET_TYPES, these are MIPS, SPARC and ALPHA now. enum sock_type in linux-user/socket.h has been extended to include TARGET_SOCK_CLOEXEC and TARGET_SOCK_NONBLOCK, similar to definition in libc. The patch also includes necessary code style changes (tab to spaces) in the header file in the MIPS #ifdef block touched by the change. Signed-off-by: Petar Jovanovic petar.jovano...@imgtec.com --- v2: - the patch defines ARCH_HAS_SOCKET_TYPES for MIPS, SPARC and ALPHA - values for sock_type are defined for SPARC and ALPHA in socket.h linux-user/socket.h | 248 +- linux-user/syscall.c | 45 + 2 files changed, 195 insertions(+), 98 deletions(-) diff --git a/linux-user/socket.h b/linux-user/socket.h index 339cae5..d2b05dc 100644 --- a/linux-user/socket.h +++ b/linux-user/socket.h @@ -1,91 +1,104 @@ #if defined(TARGET_MIPS) - // MIPS special values for constants - - /* -* For setsockopt(2) -* -* This defines are ABI conformant as far as Linux supports these ... -*/ - #define TARGET_SOL_SOCKET 0x - - #define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ - #define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ - #define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send - SIGPIPE when they die. */ - #define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ - #define TARGET_SO_BROADCAST0x0020 /* Allow transmission of - broadcast messages. */ - #define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable - socket to transmit pending data. */ - #define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. */ - #if 0 - To add: #define TARGET_SO_REUSEPORT 0x0200 /* Allow local address and port reuse. */ - #endif - - #define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ - #define TARGET_SO_STYLESO_TYPE /* Synonym */ - #define TARGET_SO_ERROR0x1007 /* get error status and clear */ - #define TARGET_SO_SNDBUF 0x1001 /* Send buffer size. */ - #define TARGET_SO_RCVBUF 0x1002 /* Receive buffer. */ - #define TARGET_SO_SNDLOWAT 0x1003 /* send low-water mark */ - #define TARGET_SO_RCVLOWAT 0x1004 /* receive low-water mark */ - #define TARGET_SO_SNDTIMEO 0x1005 /* send timeout */ - #define TARGET_SO_RCVTIMEO 0x1006 /* receive timeout */ - #define TARGET_SO_ACCEPTCONN 0x1009 - - /* linux-specific, might as well be the same as on i386 */ - #define TARGET_SO_NO_CHECK 11 - #define TARGET_SO_PRIORITY 12 - #define TARGET_SO_BSDCOMPAT14 +/* MIPS special values for constants */ + +/* + * For setsockopt(2) + * + * This defines are ABI conformant as far as Linux supports these ... + */ +#define TARGET_SOL_SOCKET 0x + +#define TARGET_SO_DEBUG0x0001 /* Record debugging information. */ +#define TARGET_SO_REUSEADDR0x0004 /* Allow reuse of local addresses. */ +#define TARGET_SO_KEEPALIVE0x0008 /* Keep connections alive and send + SIGPIPE when they die. */ +#define TARGET_SO_DONTROUTE0x0010 /* Don't do local routing. */ +#define TARGET_SO_BROADCAST0x0020 /* Allow transmission of + broadcast messages. */ +#define TARGET_SO_LINGER 0x0080 /* Block on close of a reliable +* socket to transmit pending data. +*/ +#define TARGET_SO_OOBINLINE 0x0100 /* Receive out-of-band data in-band. +*/ +#if 0 +/* To add: Allow local address and port reuse. */ +#define TARGET_SO_REUSEPORT 0x0200 +#endif + +#define TARGET_SO_TYPE 0x1008 /* Compatible name for SO_STYLE. */ +
