count: Switch from GCC to C11 thread-local storage
PATCH v2 1/1 count: Switch from GCC to C11 thread-local storage
--- CodeSamples/count/count_end.c | 12 ++++++------ count/count.tex | 20 ++++++++++++-------- 2 files changed, 18 insertions(+), 14 deletions(-)
CodeSamples/count/count_end.c +6 -6
@@ -1,6 +1,6 @@/* * count_end.c: Per-thread statistical counters that provide sum at end. -* Uses __thread for each thread's counter. +* Uses _Thread_local for each thread's counter. * * This program is free software; you can redistribute it and/or modify * it under the terms of the GNU General Public License as published by
@@ -23,17 +23,17 @@#include "../api.h" //\begin{snippet}[labelbase=ln:count:count_end:whole,commandchars=\\\@\$] -unsigned long __thread counter =0 ;//\lnlbl{var:b} +unsigned long _Thread_local counter =0 ;//\lnlbl{var:b} unsigned long *counterp[NR_THREADS ] = {NULL };unsigned long finalcount =0 ;DEFINE_SPINLOCK (final_mutex);//\lnlbl{var:e} -static __inline__void inc_count (void )//\lnlbl{inc:b} +static inline void inc_count (void )//\lnlbl{inc:b} {WRITE_ONCE (counter, counter +1 ); }//\lnlbl{inc:e} -static __inline__unsigned long read_count (void ) +static inline unsigned long read_count (void ) {int t;unsigned long sum;
@@ -48,7 +48,7 @@ static __inline__ unsigned long read_count(void) }#ifndef FCV_SNIPPET -__inline__void count_init (void ) +inline void count_init (void ) { }#endif /* FCV_SNIPPET */
@@ -73,7 +73,7 @@ void count_unregister_thread(int nthreadsexpected) //\lnlbl{unreg:b} }//\lnlbl{unreg:e} //\end{snippet} -__inline__void count_cleanup (void ) +inline void count_cleanup (void ) { }
count/count.tex +12 -8
@@ -490,7 +490,7 @@ thread (presumably cache aligned and padded to avoid false sharing). It can, and in this toy implementation, it does. But it is not that hard to come up with an alternative implementation that permits an arbitrary number of threads, - for example, using \GCC's \co{__thread} facility, + for example, using C11's \co{_Thread_local} facility, as shown in \cref{sec:count:Per-Thread-Variable-Based Implementation}. }\QuickQuizEnd
@@ -721,8 +721,12 @@ This is the topic of the next section. \subsection{Per-Thread-Variable-Based Implementation} \label{sec:count:Per-Thread-Variable-Based Implementation} -\GCC\ provides an \apig{__thread} storage class that provides -per-thread storage. +The C language, since C11, features a \apig{_Thread_local} storage class that +provides per-thread storage. +\footnote{\GCC\ provides its own \apig{__thread} storage class, which was used +in previous versions of this book. +The two methods for specifying a thread-local variable are interchangeable +when using \GCC\@.} This can be used as shown in \cref{lst:count:Per-Thread Statistical Counters} (\path{count_end.c}) to implement
@@ -749,14 +753,14 @@ value of the counter and exiting threads. Doesn't that explicit \co{counterp} array in \cref{lst:count:Per-Thread Statistical Counters} reimpose an arbitrary limit on the number of threads? - Why doesn't \GCC\ provide a \co{per_thread()} interface, similar + Why doesn't the C language provide a \co{per_thread()} interface, similar to the Linux kernel's \co{per_cpu()} primitive, to allow threads to more easily access each others' per-thread variables? }\QuickQuizAnswer{ Why indeed? - To be fair, \GCC\ faces some challenges that the Linux kernel - gets to ignore. + To be fair, user-mode thread-local storage faces some challenges + that the Linux kernel gets to ignore. When a user-level thread exits, its per-thread variables all disappear, which complicates the problem of per-thread-variable access, particularly before the advent of user-level RCU
@@ -940,7 +944,7 @@ variables vanish when that thread exits. more graceful manner. }\QuickQuizEnd -Both the array-based and \co{__thread}-based approaches offer excellent +Both the array-based and \co{_Thread_local}-based approaches offer excellent update-side performance and scalability. However, these benefits result in large read-side expense for large numbers of threads.
@@ -2961,7 +2965,7 @@ courtesy of eventual consistency. work when there are more threads. In addition, the last two algorithms interpose an additional level of indirection because they map from integer thread ID - to the corresponding \co{__thread} variable. + to the corresponding \co{_Thread_local} variable. }\QuickQuizEndB % \QuickQuizE{
Signature · 1 line
2.25.1
Hi Elad,
Quoted 2 lines · Elad Lahav
Signed-off-by Elad Lahav <[email protected]>---
As there is no changelog here, let me ask a (maybe stupid) question.
What is your goal of diverting from GCC extensions and switching to the C11 standard?
Do you want the codebase under CodeSamples/ to be strictly conformant to C11 or later?
Runnig "make" under CodeSamples/count, with "-std=c11" appended to GCC_ARGS, I get a lot of compile errors/warnings (with GCC 9.4.0 under Ubuntu 20.04), beginning with:
cc -g -O2 -Wall -std=c11 -o count_atomic count_atomic.c -lpthread
In file included from /usr/include/sched.h:34,
from /usr/include/pthread.h:22,
from ../api.h:159,
from count_atomic.c:22:
/usr/include/time.h:113:5: error: unknown type name 'locale_t'
113 | locale_t __loc) __THROW;
| ^~~~~~~~A workaround is to append "-D_GNU_SOURCE" to GCC_ARGS. With that, I get the next warning of:
../api.h:766:2: warning: implicit declaration of function 'typeof' [-Wimplicit-function-declaration]
766 | typeof(*ptr) _____actual = (o); \
Apparently, typeof() is another GCC extension. __typeof__() might be used instead, but it is not ISO C.
So what is you goal of these switches?
Thanks, Akira
Quoted 4 lines
CodeSamples/count/count_end.c | 12 ++++++------ count/count.tex | 20 ++++++++++++-------- 2 files changed, 18 insertions(+), 14 deletions(-)
[...]
Hi Akira,
The reason for the change is that using C11 is more portable. I have no problem dropping it if it causes any problems.
That said, there is a simple solution to the issue you are facing. The -std=c11 option specifies strict adherence to ISO C, without extensions. You rarely want that. Instead, specify -std=gnu11, which will give you C11 plus POSIX plus GNU extensions.
--Elad
Quoted 57 lines · On Wed., Aug. 17, 2022, 11:18 a.m. Akira Yokosawa, <[email protected]> wrote:
Hi Elad,
Quoted 2 lines · Elad Lahav
Signed-off-by Elad Lahav <[email protected]>---
As there is no changelog here, let me ask a (maybe stupid) question.
What is your goal of diverting from GCC extensions and switching to the C11 standard?
Do you want the codebase under CodeSamples/ to be strictly conformant to C11 or later?
Runnig "make" under CodeSamples/count, with "-std=c11" appended to
GCC_ARGS, I get a lot of compile errors/warnings (with GCC 9.4.0 under Ubuntu 20.04), beginning with:cc -g -O2 -Wall -std=c11 -o count_atomic count_atomic.c -lpthread In file included from /usr/include/sched.h:34, from /usr/include/pthread.h:22, from ../api.h:159, from count_atomic.c:22: /usr/include/time.h:113:5: error: unknown type name 'locale_t' 113 | locale_t __loc) __THROW; | ^~~~~~~~A workaround is to append "-D_GNU_SOURCE" to
GCC_ARGS. With that, I get the next warning of:
../api.h:766:2: warning: implicit declaration of function 'typeof'
[-Wimplicit-function-declaration]766 | typeof(*ptr) _____actual = (o); \Apparently,
typeof()is another GCC extension.__typeof__()might be used instead, but it is not ISO C.So what is you goal of these switches?
Thanks, Akira
Quoted 4 lines
CodeSamples/count/count_end.c | 12 ++++++------ count/count.tex | 20 ++++++++++++-------- 2 files changed, 18 insertions(+), 14 deletions(-)[...]
Incrementalist that I am, my thought is to take this one thing at a time, if and when. ;-0
Thanx, Paul
Quoted 71 lines · Elad Lahav
Hi Akira,
The reason for the change is that using C11 is more portable. I have no problem dropping it if it causes any problems.
That said, there is a simple solution to the issue you are facing. The -std=c11 option specifies strict adherence to ISO C, without extensions. You rarely want that. Instead, specify -std=gnu11, which will give you C11 plus POSIX plus GNU extensions.
--Elad
Quoted 57 lines · On Wed., Aug. 17, 2022, 11:18 a.m. Akira Yokosawa, <[email protected]> wrote:
Hi Elad,
Quoted 2 lines · Elad Lahav
Signed-off-by Elad Lahav <[email protected]>---
As there is no changelog here, let me ask a (maybe stupid) question.
What is your goal of diverting from GCC extensions and switching to the C11 standard?
Do you want the codebase under CodeSamples/ to be strictly conformant to C11 or later?
Runnig "make" under CodeSamples/count, with "-std=c11" appended to
GCC_ARGS, I get a lot of compile errors/warnings (with GCC 9.4.0 under Ubuntu 20.04), beginning with:cc -g -O2 -Wall -std=c11 -o count_atomic count_atomic.c -lpthread In file included from /usr/include/sched.h:34, from /usr/include/pthread.h:22, from ../api.h:159, from count_atomic.c:22: /usr/include/time.h:113:5: error: unknown type name 'locale_t' 113 | locale_t __loc) __THROW; | ^~~~~~~~A workaround is to append "-D_GNU_SOURCE" to
GCC_ARGS. With that, I get the next warning of:
../api.h:766:2: warning: implicit declaration of function 'typeof'
[-Wimplicit-function-declaration]766 | typeof(*ptr) _____actual = (o); \Apparently,
typeof()is another GCC extension.__typeof__()might be used instead, but it is not ISO C.So what is you goal of these switches?
Thanks, Akira
Quoted 4 lines
CodeSamples/count/count_end.c | 12 ++++++------ count/count.tex | 20 ++++++++++++-------- 2 files changed, 18 insertions(+), 14 deletions(-)[...]
Quoted 2 lines · Paul E. McKenney
Incrementalist that I am, my thought is to take this one thing at a time, if and when. ;-0
What would you like to do? I have no objection to dropping this change, but I think (at least in principle) that using standard language features is better than using compiler-specific extensions. If nothing else it makes the MISRA folks happy ;-)
--Elad
Paul E. McKenney
AppliedQuoted 8 lines · Elad Lahav
Quoted 2 lines · Paul E. McKenney
Incrementalist that I am, my thought is to take this one thing at a time, if and when. ;-0
What would you like to do? I have no objection to dropping this change, but I think (at least in principle) that using standard language features is better than using compiler-specific extensions. If nothing else it makes the MISRA folks happy ;-)
;-)
Applied, thank you both!
Thanx, Paul