Return-Path: DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20161025; h=subject:to:cc:references:from:message-id:date:user-agent :mime-version:in-reply-to:content-language:content-transfer-encoding; bh=1HG5DIaqLf9w7M9csBEHiP3rtLzwLKPsM3Ewfvm0Ujw=; b=H1FpdTxZPfOnptTOUbB9G7IozLsDp5H0DYL5+XKFR+i5nXX+yP336BVsFVOQ5Rau1+ 4bovpWij3ivKuoz0vLmcGnamyI5wsoGCH7HUUHpiz7iloCp4tsrKCwh0V9RJj4NbFIZc m5unRQVHBBNcYJUiE3qdTjl1v7vnKRQTn5pya9hgH/Z+Tbu6S6E2wQNFZYb5wXqRXDTZ 8sHtQrk2H8gIswQgYwZ0EupOZoVjiraijDDraFqP34Asze7KGp6zzrjDUGgrjzLu3H88 9WZ0+80mBHbEFvPWozmz4Ml+HQyr1H3dwO4vegphiVPmVaJDAMRagmHZXu/JTWLV49HQ +BnA== Subject: [PATCH 2/7] toolsoftrade: Use 'VerbatimU' for inline snippets References: From: Akira Yokosawa Message-ID: <7bdd087c-bb71-56d8-e32c-9b000faf3aec@gmail.com> Date: Sat, 27 Oct 2018 00:24:52 +0900 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit To: "Paul E. McKenney" Cc: perfbook@vger.kernel.org, Akira Yokosawa List-ID: From f65277139c063f4b44987edb0eee40ade1957df2 Mon Sep 17 00:00:00 2001 From: Akira Yokosawa Date: Sun, 14 Oct 2018 14:54:14 +0900 Subject: [PATCH 2/7] toolsoftrade: Use 'VerbatimU' for inline snippets Signed-off-by: Akira Yokosawa --- toolsoftrade/toolsoftrade.tex | 90 ++++++++++--------------------------------- 1 file changed, 20 insertions(+), 70 deletions(-) diff --git a/toolsoftrade/toolsoftrade.tex b/toolsoftrade/toolsoftrade.tex index 6896997..601829c 100644 --- a/toolsoftrade/toolsoftrade.tex +++ b/toolsoftrade/toolsoftrade.tex @@ -88,14 +88,9 @@ of \co{cat} execute sequentially. \QuickQuizAnswer{ One straightforward approach is the shell pipeline: -\begin{minipage}[t]{\columnwidth} -\vspace{0.1ex} -\small -\begin{verbatim} +\begin{VerbatimU} grep $pattern1 | sed -e 's/a/b/' | sort -\end{verbatim} -\vspace{0.1ex} -\end{minipage} +\end{VerbatimU} For a sufficiently large input file, \co{grep} will pattern-match in parallel with \co{sed} @@ -648,15 +643,10 @@ Lines~7-11 create a thread running \co{lock_writer()}. Lines~12-19 wait for both threads to complete. The output of this code fragment is as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\scriptsize -\begin{verbatim} +\begin{VerbatimU} Creating two threads using same lock: lock_reader(): x = 0 -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} Because both threads are using the same lock, the \co{lock_reader()} thread cannot see any of the intermediate values of \co{x} produced @@ -718,18 +708,13 @@ shows a similar code fragment, but this time using different locks: \co{lock_writer()}. The output of this code fragment is as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\scriptsize -\begin{verbatim} +\begin{VerbatimU} Creating two threads w/different locks: lock_reader(): x = 0 lock_reader(): x = 1 lock_reader(): x = 2 lock_reader(): x = 3 -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} Because the two threads are using different locks, they do not exclude each other, and can run concurrently. @@ -1144,16 +1129,11 @@ If you instead need the new value, you can instead use the For example, one could implement \co{__sync_nand_and_fetch()} in terms of \co{__sync_fetch_and_nand()} as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\scriptsize -\begin{verbatim} +\begin{VerbatimU} tmp = v; ret = __sync_fetch_and_nand(p, tmp); ret = ~ret & tmp; -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} It is similarly possible to implement \co{__sync_fetch_and_add()}, \co{__sync_fetch_and_sub()}, and \co{__sync_fetch_and_xor()} @@ -1615,16 +1595,11 @@ allowing other threads to acquire it. A spinlock named \co{mutex} may be used to protect a variable \co{counter} as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\small -\begin{verbatim} +\begin{VerbatimU} spin_lock(&mutex); counter++; spin_unlock(&mutex); -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} \QuickQuiz{} What problems could occur if the variable {\tt counter} were @@ -1633,16 +1608,11 @@ spin_unlock(&mutex); On CPUs with load-store architectures, incrementing {\tt counter} might compile into something like the following: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\small -\begin{verbatim} +\begin{VerbatimU} LOAD counter,r0 INC r0 STORE r0,counter -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} On such machines, two threads might simultaneously load the value of {\tt counter}, each increment it, and each store the @@ -2505,51 +2475,31 @@ Section~\ref{sec:count:Statistical Counters}, it is helpful to implement such a counter using a per-thread variable. Such a variable can be defined as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\small -\begin{verbatim} +\begin{VerbatimU} DEFINE_PER_THREAD(int, counter); -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} The counter must be initialized as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\small -\begin{verbatim} +\begin{VerbatimU} init_per_thread(counter, 0); -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} A thread can increment its instance of this counter as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\small -\begin{verbatim} +\begin{VerbatimU} p_counter = &__get_thread_var(counter); WRITE_ONCE(*p_counter, *p_counter + 1); -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} The value of the counter is then the sum of its instances. A snapshot of the value of the counter can thus be collected as follows: -\vspace{5pt} -\begin{minipage}[t]{\columnwidth} -\small -\begin{verbatim} +\begin{VerbatimU} for_each_thread(t) sum += READ_ONCE(per_thread(counter, t)); -\end{verbatim} -\end{minipage} -\vspace{5pt} +\end{VerbatimU} Again, it is possible to gain a similar effect using other mechanisms, but per-thread variables combine convenience and high performance, -- 2.7.4