From: Adhemerval Zanella Netto <adhemerval.zanella@linaro.org>
To: DJ Delorie <dj@redhat.com>
Cc: libc-alpha@sourceware.org
Subject: Re: [PATCH] malloc: Use C11 like atomics on memusage
Date: Thu, 23 Feb 2023 13:39:05 -0300 [thread overview]
Message-ID: <2a430a50-4c14-d986-014f-8c816f105f52@linaro.org> (raw)
In-Reply-To: <xnedqh8123.fsf@greed.delorie.com>
Hi D.J,
Wilco has objected that these should be relaxed MO [1], so I plan to
send an updated version to fix it.
[1] https://sourceware.org/pipermail/libc-alpha/2023-February/145665.html
On 23/02/23 01:20, DJ Delorie wrote:
>
> LGTM.
>
> Reviewed-by: DJ Delorie <dj@redhat.com>
>
> Adhemerval Zanella via Libc-alpha <libc-alpha@sourceware.org> writes:
>> +static inline void
>> +peak_atomic_max (size_t *peak, size_t val)
>> +{
>> + size_t v;
>> + do
>> + {
>> + v = atomic_load_relaxed (peak);
>> + if (v >= val)
>> + break;
>> + }
>> + while (! atomic_compare_exchange_weak_acquire (peak, &v, val));
>> +}
>> +
>
> This is the only call without a direct replacement. This inline
> replicates what <atomic.h> does. Ok.
>
>> - = catomic_exchange_and_add (¤t_heap, len - old_len) + len - old_len;
>> - catomic_max (&peak_heap, heap);
>> + = atomic_fetch_add_acquire (¤t_heap, len - old_len) + len - old_len;
>> + peak_atomic_max (&peak_heap, heap);
>
> Ok.
>
>> - catomic_max (&peak_stack, current_stack);
>> + peak_atomic_max (&peak_stack, current_stack);
>
> Ok.
>
>> - catomic_max (&peak_total, heap + current_stack);
>> + peak_atomic_max (&peak_total, heap + current_stack);
>
> Ok.
>
>> - uint32_t idx = catomic_exchange_and_add (&buffer_cnt, 1);
>> + uint32_t idx = atomic_fetch_add_acquire (&buffer_cnt, 1);
>
> Ok.
>
>> - catomic_compare_and_exchange_val_acq (&buffer_cnt, reset, idx + 1);
>> + uint32_t expected = idx + 1;
>> + atomic_compare_exchange_weak_acquire (&buffer_cnt, &expected, reset);
>
> Ok.
>
>> - catomic_increment (&calls[idx_malloc]);
>> + atomic_fetch_add_acquire (&calls[idx_malloc], 1);
>
> Ok.
>
>> - catomic_add (&total[idx_malloc], len);
>> + atomic_fetch_add_acquire (&total[idx_malloc], len);
>
> Ok.
>
>> - catomic_add (&grand_total, len);
>> + atomic_fetch_add_acquire (&grand_total, len);
>
> Ok.
>
>> if (len < 65536)
>> - catomic_increment (&histogram[len / 16]);
>> + atomic_fetch_add_acquire (&histogram[len / 16], 1);
>> else
>> - catomic_increment (&large);
>> + atomic_fetch_add_acquire (&large, 1);
>
> Ok.
>
>> - catomic_increment (&calls_total);
>> + atomic_fetch_add_acquire (&calls_total, 1);
>
> Ok.
>
>> - catomic_increment (&failed[idx_malloc]);
>> + atomic_fetch_add_acquire (&failed[idx_malloc], 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx_realloc]);
>> + atomic_fetch_add_acquire (&calls[idx_realloc], 1);
>
> Ok.
>
>> - catomic_add (&total[idx_realloc], len - old_len);
>> + atomic_fetch_add_acquire (&total[idx_realloc], len - old_len);
>
> Ok.
>
>> - catomic_add (&grand_total, len - old_len);
>> + atomic_fetch_add_acquire (&grand_total, len - old_len);
>
> Ok.
>
>> - catomic_increment (&realloc_free);
>> + atomic_fetch_add_acquire (&realloc_free, 1);
>
> Ok.
>
>> - catomic_add (&total[idx_free], real->length);
>> + atomic_fetch_add_acquire (&total[idx_free], real->length);
>
> Ok.
>
>> if (len < 65536)
>> - catomic_increment (&histogram[len / 16]);
>> + atomic_fetch_add_acquire (&histogram[len / 16], 1);
>> else
>> - catomic_increment (&large);
>> + atomic_fetch_add_acquire (&large, 1);
>
> Ok.
>
>> - catomic_increment (&calls_total);
>> + atomic_fetch_add_acquire (&calls_total, 1);
>
> Ok.
>
>> - catomic_increment (&failed[idx_realloc]);
>> + atomic_fetch_add_acquire (&failed[idx_realloc], 1);
>
> Ok.
>
>> - catomic_increment (&inplace);
>> + atomic_fetch_add_acquire (&inplace, 1);
>
> Ok.
>
>> - catomic_increment (&decreasing);
>> + atomic_fetch_add_acquire (&decreasing, 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx_calloc]);
>> + atomic_fetch_add_acquire (&calls[idx_calloc], 1);
>
> Ok.
>
>> - catomic_add (&total[idx_calloc], size);
>> + atomic_fetch_add_acquire (&total[idx_calloc], size);
>
> Ok.
>
>> - catomic_add (&grand_total, size);
>> + atomic_fetch_add_acquire (&grand_total, size);
>
> Ok.
>
>> if (size < 65536)
>> - catomic_increment (&histogram[size / 16]);
>> + atomic_fetch_add_acquire (&histogram[size / 16], 1);
>> else
>> - catomic_increment (&large);
>> + atomic_fetch_add_acquire (&large, 1);
>
> Ok.
>
>> - catomic_increment (&failed[idx_calloc]);
>> + atomic_fetch_add_acquire (&failed[idx_calloc], 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx_free]);
>> + atomic_fetch_add_acquire (&calls[idx_free], 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx_free]);
>> + atomic_fetch_add_acquire (&calls[idx_free], 1);
>
> Ok.
>
>> - catomic_add (&total[idx_free], real->length);
>> + atomic_fetch_add_acquire (&total[idx_free], real->length);
>
> Ok.
>
>> - catomic_increment (&calls[idx]);
>> + atomic_fetch_add_acquire (&calls[idx], 1);
>
> Ok.
>
>> - catomic_add (&total[idx], len);
>> + atomic_fetch_add_acquire (&total[idx], len);
>
> Ok.
>
>> - catomic_add (&grand_total, len);
>> + atomic_fetch_add_acquire (&grand_total, len);
>
> Ok.
>
>> if (len < 65536)
>> - catomic_increment (&histogram[len / 16]);
>> + atomic_fetch_add_acquire (&histogram[len / 16], 1);
>> else
>> - catomic_increment (&large);
>> + atomic_fetch_add_acquire (&large, 1);
>
> Ok.
>
>> - catomic_increment (&calls_total);
>> + atomic_fetch_add_acquire (&calls_total, 1);
>
> Ok.
>
>> - catomic_increment (&failed[idx]);
>> + atomic_fetch_add_acquire (&failed[idx], 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx]);
>> + atomic_fetch_add_acquire (&calls[idx], 1);
>
> Ok.
>
>> - catomic_add (&total[idx], len);
>> + atomic_fetch_add_acquire (&total[idx], len);
>
> Ok.
>
>> - catomic_add (&grand_total, len);
>> + atomic_fetch_add_acquire (&grand_total, len);
>
> Ok.
>
>> if (len < 65536)
>> - catomic_increment (&histogram[len / 16]);
>> + atomic_fetch_add_acquire (&histogram[len / 16], 1);
>> else
>> - catomic_increment (&large);
>> + atomic_fetch_add_acquire (&large, 1);
>
> Ok.
>
>> - catomic_increment (&calls_total);
>> + atomic_fetch_add_acquire (&calls_total, 1);
>
> Ok.
>
>> - catomic_increment (&failed[idx]);
>> + atomic_fetch_add_acquire (&failed[idx], 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx_mremap]);
>> + atomic_fetch_add_acquire (&calls[idx_mremap], 1);
>
> Ok.
>
>> - catomic_add (&total[idx_mremap], len - old_len);
>> + atomic_fetch_add_acquire (&total[idx_mremap], len - old_len);
>
> Ok.
>
>> - catomic_add (&grand_total, len - old_len);
>> + atomic_fetch_add_acquire (&grand_total, len - old_len);
>
> Ok.
>
>
>> if (len < 65536)
>> - catomic_increment (&histogram[len / 16]);
>> + atomic_fetch_add_acquire (&histogram[len / 16], 1);
>> else
>> - catomic_increment (&large);
>> + atomic_fetch_add_acquire (&large, 1);
>
> Ok.
>
>> - catomic_increment (&calls_total);
>> + atomic_fetch_add_acquire (&calls_total, 1);
>
> Ok.
>
>> - catomic_increment (&failed[idx_mremap]);
>> + atomic_fetch_add_acquire (&failed[idx_mremap], 1);
>
> Ok.
>
>> - catomic_increment (&inplace_mremap);
>> + atomic_fetch_add_acquire (&inplace_mremap, 1);
>
> Ok.
>
>> - catomic_increment (&decreasing_mremap);
>> + atomic_fetch_add_acquire (&decreasing_mremap, 1);
>
> Ok.
>
>> - catomic_increment (&calls[idx_munmap]);
>> + atomic_fetch_add_acquire (&calls[idx_munmap], 1);
>
> Ok.
>
>> - catomic_add (&total[idx_munmap], len);
>> + atomic_fetch_add_acquire (&total[idx_munmap], len);
>
> Ok.
>
>> - catomic_increment (&failed[idx_munmap]);
>> + atomic_fetch_add_acquire (&failed[idx_munmap], 1);
>
> Ok.
>
next prev parent reply other threads:[~2023-02-23 16:39 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-31 18:14 Adhemerval Zanella
2023-02-14 18:47 ` Adhemerval Zanella Netto
2023-02-23 4:20 ` DJ Delorie
2023-02-23 16:39 ` Adhemerval Zanella Netto [this message]
2023-02-23 19:17 ` DJ Delorie
2023-02-16 18:27 Wilco Dijkstra
2023-02-16 18:39 ` Adhemerval Zanella Netto
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=2a430a50-4c14-d986-014f-8c816f105f52@linaro.org \
--to=adhemerval.zanella@linaro.org \
--cc=dj@redhat.com \
--cc=libc-alpha@sourceware.org \
/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;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).