From: nils.carlson@ludd.ltu.se (Nils Carlson)
Subject: [ltt-dev] [RFC UST PATCH] Put back urcu_ref counter in buffers.c
Date: Mon, 14 Feb 2011 20:57:13 +0100 [thread overview]
Message-ID: <77FC3B3C-1E96-467D-90FA-08E9218AFD49@ludd.ltu.se> (raw)
In-Reply-To: <1297704926-17478-1-git-send-email-david.goulet@polymtl.ca>
Hi David,
Your patch says what it does, but nothing about why. Please elaborate.
/Nils
On Feb 14, 2011, at 6:35 PM, David Goulet wrote:
> Signed-off-by: David Goulet <david.goulet at polymtl.ca>
> ---
> libust/buffers.c | 133 +++++++++++++++++++++++++++++++++
> +-------------------
> 1 files changed, 85 insertions(+), 48 deletions(-)
>
> diff --git a/libust/buffers.c b/libust/buffers.c
> index 4e8004c..a82538b 100644
> --- a/libust/buffers.c
> +++ b/libust/buffers.c
> @@ -28,6 +28,7 @@
> #include <stdlib.h>
>
> #include <ust/clock.h>
> +#include <urcu/urcu_ref.h>
>
> #include "buffers.h"
> #include "channels.h"
> @@ -150,6 +151,18 @@ static void ltt_buffer_begin(struct ust_buffer
> *buf,
> ltt_write_trace_header(channel->trace, header);
> }
>
> +static int unmap_buf_structs(struct ust_channel *chan)
> +{
> + int i;
> +
> + for (i=0; i < chan->n_cpus; i++) {
> + if (shmdt(chan->buf[i]) < 0) {
> + PERROR("shmdt");
> + }
> + }
> + return 0;
> +}
> +
> static int map_buf_data(struct ust_buffer *buf, size_t *size)
> {
> void *ptr;
> @@ -208,11 +221,17 @@ static int open_buf(struct ust_channel *chan,
> int cpu)
> if (result < 0)
> return -1;
>
> + urcu_ref_init(&chan->buf[cpu]->urcu_ref);
> +
> buf->commit_count =
> zmalloc(sizeof(*buf->commit_count) * n_subbufs);
> if (!buf->commit_count)
> goto unmap_buf;
>
> + urcu_ref_get(&trace->urcu_ref);
> + urcu_ref_get(&trace->ltt_transport_urcu_ref);
> + urcu_ref_get(&chan->urcu_ref);
> +
> result = pipe(fds);
> if (result < 0) {
> PERROR("pipe");
> @@ -222,7 +241,9 @@ static int open_buf(struct ust_channel *chan,
> int cpu)
> buf->data_ready_fd_write = fds[1];
>
> buf->cpu = cpu;
> +
> buf->chan = chan;
> + urcu_ref_get(&chan->urcu_ref);
>
> uatomic_set(&buf->offset, ltt_subbuffer_header_size());
> uatomic_set(&buf->consumed, 0);
> @@ -254,13 +275,17 @@ unmap_buf:
> return -1;
> }
>
> +static void release_channel(struct urcu_ref *);
> static void ltt_relay_print_buffer_errors(struct ust_channel *chan,
> int cpu);
>
> -static void close_buf(struct ust_buffer *buf)
> +static void destroy_buf(struct ust_buffer *buf)
> {
> + int result;
> struct ust_channel *chan = buf->chan;
> int cpu = buf->cpu;
> - int result;
> +
> + /* Print possible buffer errors */
> + ltt_relay_print_buffer_errors(chan, cpu);
>
> result = shmdt(buf->buf_data);
> if (result < 0) {
> @@ -279,10 +304,61 @@ static void close_buf(struct ust_buffer *buf)
> PERROR("close");
> }
>
> - /* FIXME: This spews out errors, are they real?:
> - * ltt_relay_print_buffer_errors(chan, cpu); */
> + urcu_ref_put(&chan->urcu_ref, release_channel);
> + urcu_ref_put(&chan->trace->urcu_ref, ltt_release_trace);
> }
>
> +static void remove_buf(struct urcu_ref *urcu_ref)
> +{
> + struct ust_buffer *buf = _ust_container_of(urcu_ref, struct
> ust_buffer, urcu_ref);
> + destroy_buf(buf);
> +}
> +
> +static void close_buf(struct ust_buffer *buf)
> +{
> + urcu_ref_put(&buf->urcu_ref, remove_buf);
> +}
> +
> +static void destroy_channel(struct urcu_ref *urcu_ref)
> +{
> + struct ust_channel *chan = _ust_container_of(urcu_ref, struct
> ust_channel, urcu_ref);
> +
> + free(chan);
> +}
> +
> +static void release_channel(struct urcu_ref *urcu_ref)
> +{
> + struct ust_channel *chan = _ust_container_of(urcu_ref,
> + struct ust_channel, urcu_ref);
> +
> + free(chan->buf);
> +}
> +
> +static void close_channel(struct ust_channel *chan)
> +{
> + int i;
> + if(!chan)
> + return;
> +
> + pthread_mutex_lock(&ust_buffers_channels_mutex);
> + for(i=0; i<chan->n_cpus; i++) {
> + /* FIXME: if we make it here, then all buffers were necessarily
> allocated. Moreover, we don't
> + * initialize to NULL so we cannot use this check. Should we? */
> + //ust// if (chan->buf[i])
> + close_buf(chan->buf[i]);
> + }
> +
> + cds_list_del(&chan->list);
> + urcu_ref_put(&chan->urcu_ref, destroy_channel);
> + pthread_mutex_unlock(&ust_buffers_channels_mutex);
> +}
> +
> +static void remove_channel(struct ust_channel *chan)
> +{
> + close_channel(chan);
> + unmap_buf_structs(chan);
> + urcu_ref_put(&chan->urcu_ref, release_channel);
> +}
>
> static int open_channel(struct ust_channel *chan, size_t subbuf_size,
> size_t subbuf_cnt)
> @@ -308,6 +384,8 @@ static int open_channel(struct ust_channel
> *chan, size_t subbuf_size,
> chan->subbuf_size_order = get_count_order(subbuf_size);
> chan->alloc_size = subbuf_size * subbuf_cnt;
>
> + urcu_ref_init(&chan->urcu_ref);
> +
> pthread_mutex_lock(&ust_buffers_channels_mutex);
> for (i=0; i < chan->n_cpus; i++) {
> result = open_buf(chan, i);
> @@ -327,29 +405,11 @@ error:
> do {} while(0);
> }
>
> + urcu_ref_put(&chan->urcu_ref, destroy_channel);
> pthread_mutex_unlock(&ust_buffers_channels_mutex);
> return -1;
> }
>
> -static void close_channel(struct ust_channel *chan)
> -{
> - int i;
> - if(!chan)
> - return;
> -
> - pthread_mutex_lock(&ust_buffers_channels_mutex);
> - for(i=0; i<chan->n_cpus; i++) {
> - /* FIXME: if we make it here, then all buffers were necessarily
> allocated. Moreover, we don't
> - * initialize to NULL so we cannot use this check. Should we? */
> -//ust// if (chan->buf[i])
> - close_buf(chan->buf[i]);
> - }
> -
> - cds_list_del(&chan->list);
> -
> - pthread_mutex_unlock(&ust_buffers_channels_mutex);
> -}
> -
> static void ltt_force_switch(struct ust_buffer *buf,
> enum force_switch_mode mode);
>
> @@ -646,18 +706,6 @@ static int map_buf_structs(struct ust_channel
> *chan)
> return -1;
> }
>
> -static int unmap_buf_structs(struct ust_channel *chan)
> -{
> - int i;
> -
> - for (i=0; i < chan->n_cpus; i++) {
> - if (shmdt(chan->buf[i]) < 0) {
> - PERROR("shmdt");
> - }
> - }
> - return 0;
> -}
> -
> /*
> * Create channel.
> */
> @@ -667,6 +715,8 @@ static int create_channel(const char
> *trace_name, struct ust_trace *trace,
> {
> int i, result;
>
> + urcu_ref_init(&chan->urcu_ref);
> +
> chan->trace = trace;
> chan->overwrite = overwrite;
> chan->n_subbufs_order = get_count_order(n_subbufs);
> @@ -714,19 +764,6 @@ error:
> return -1;
> }
>
> -
> -static void remove_channel(struct ust_channel *chan)
> -{
> - close_channel(chan);
> -
> - unmap_buf_structs(chan);
> -
> - free(chan->buf_struct_shmids);
> -
> - free(chan->buf);
> -
> -}
> -
> static void ltt_relay_async_wakeup_chan(struct ust_channel
> *ltt_channel)
> {
> //ust// unsigned int i;
> --
> 1.7.4
>
>
> _______________________________________________
> ltt-dev mailing list
> ltt-dev at lists.casi.polymtl.ca
> http://lists.casi.polymtl.ca/cgi-bin/mailman/listinfo/ltt-dev
prev parent reply other threads:[~2011-02-14 19:57 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-02-14 17:35 David Goulet
2011-02-14 19:57 ` Nils Carlson [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=77FC3B3C-1E96-467D-90FA-08E9218AFD49@ludd.ltu.se \
--to=nils.carlson@ludd.ltu.se \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox