From: Ben Widawsky <benjamin.widawsky@intel.com>
To: Intel GFX <intel-gfx@lists.freedesktop.org>
Cc: Ben Widawsky <ben@bwidawsk.net>,
Ben Widawsky <benjamin.widawsky@intel.com>
Subject: [PATCH 4/9] drm/i915: Limit the number of node allocation retries
Date: Tue, 6 May 2014 22:21:33 -0700 [thread overview]
Message-ID: <1399440098-17378-4-git-send-email-benjamin.widawsky@intel.com> (raw)
In-Reply-To: <1399440098-17378-1-git-send-email-benjamin.widawsky@intel.com>
AFAICT, it's impossible to actually infinitely retry the allocation in
our current code. However, a small oversight on my part, slight bug, or
future bug, could easily change that.
To avoid a potential breakage, a simple retry variable of 16 bits will
help us prevent infinitely running.
Retry is limited to 100 as a mostly random number. Some consideration
about stack usage was taken into account. As an example, if we allowed
256 retries on a 32b arch (and my memory serves that all arguments are
passed on the stack for such architectures), thats 33 bytes * 256
retries + (fudge for return address and such)... it's way more than we
want to use already. 64b architectures might be slightly better, since
6? of the first args will get passed through registers, but it's still
bad.
If anything, we might want to do way less than 100, like 3.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
---
drivers/gpu/drm/i915/i915_gem.c | 14 +++++++++-----
1 file changed, 9 insertions(+), 5 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c
index b6965a2..de98b26 100644
--- a/drivers/gpu/drm/i915/i915_gem.c
+++ b/drivers/gpu/drm/i915/i915_gem.c
@@ -3216,6 +3216,7 @@ static void i915_gem_verify_gtt(struct drm_device *dev)
#endif
}
+#define MAX_VMA_FIND_RETRY 100
static int
i915_gem_find_vm_space(struct i915_address_space *vm,
struct drm_mm_node *node,
@@ -3224,7 +3225,8 @@ i915_gem_find_vm_space(struct i915_address_space *vm,
unsigned long color,
unsigned long start,
unsigned long end,
- uint32_t flags)
+ uint32_t flags,
+ uint8_t retry)
{
int ret;
ret = drm_mm_insert_node_in_range_generic(&vm->mm, node,
@@ -3232,7 +3234,7 @@ i915_gem_find_vm_space(struct i915_address_space *vm,
start, end,
DRM_MM_SEARCH_DEFAULT,
DRM_MM_CREATE_DEFAULT);
- if (ret) {
+ if (ret && (retry < MAX_VMA_FIND_RETRY)) {
if (WARN_ON(ret != -ENOSPC))
return ret;
@@ -3241,7 +3243,8 @@ i915_gem_find_vm_space(struct i915_address_space *vm,
if (ret == 0)
return i915_gem_find_vm_space(vm, node,
size, align, color,
- start, end, flags);
+ start, end, flags,
+ retry++);
}
return ret;
@@ -3306,8 +3309,9 @@ i915_gem_object_bind_to_vm(struct drm_i915_gem_object *obj,
if (IS_ERR(vma))
goto err_unpin;
- ret = i915_gem_find_vm_space(vm, &vma->node, size, alignment,
- obj->cache_level, 0, gtt_max, flags);
+ ret = i915_gem_find_vm_space(vm, &vma->node,
+ size, alignment, obj->cache_level,
+ 0, gtt_max, flags, 1);
if (ret)
goto err_free_vma;
--
1.9.2
next prev parent reply other threads:[~2014-05-07 5:21 UTC|newest]
Thread overview: 23+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-05-07 5:21 [PATCH 1/9] drm/i915: Use topdown allocation for PPGTT PDEs on gen6/7 Ben Widawsky
2014-05-07 5:21 ` [PATCH 2/9] drm/i915: Extract node allocation from bind Ben Widawsky
2014-05-07 7:02 ` Chris Wilson
2014-05-07 15:45 ` Ben Widawsky
2014-05-07 15:53 ` Chris Wilson
2014-05-07 16:00 ` Ben Widawsky
2014-05-07 16:55 ` Chris Wilson
2014-05-07 17:30 ` Ben Widawsky
2014-05-07 5:21 ` [PATCH 3/9] drm/i915: WARN on unexpected return from drm_mm Ben Widawsky
2014-05-07 5:21 ` Ben Widawsky [this message]
2014-05-07 7:49 ` [PATCH 4/9] drm/i915: Limit the number of node allocation retries Daniel Vetter
2014-05-07 15:21 ` Ben Widawsky
2014-05-07 5:21 ` [PATCH 5/9] drm/i915: Use new drm node allocator for PPGTT Ben Widawsky
2014-05-07 5:21 ` [PATCH 6/9] drm/i915: Wrap VMA binding Ben Widawsky
2014-05-07 7:55 ` Daniel Vetter
2014-05-07 15:54 ` Ben Widawsky
2014-05-07 16:09 ` Daniel Vetter
2014-05-07 5:21 ` [PATCH 7/9] drm/i915: Make aliasing a 2nd class VM Ben Widawsky
2014-05-07 7:56 ` Daniel Vetter
2014-05-07 5:21 ` [PATCH 8/9] drm/i915: Make pin global flags explicit Ben Widawsky
2014-05-07 5:21 ` [PATCH 9/9] drm/i915: Split out aliasing binds Ben Widawsky
2014-05-07 7:59 ` Daniel Vetter
2014-05-07 7:44 ` [PATCH 1/9] drm/i915: Use topdown allocation for PPGTT PDEs on gen6/7 Daniel Vetter
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=1399440098-17378-4-git-send-email-benjamin.widawsky@intel.com \
--to=benjamin.widawsky@intel.com \
--cc=ben@bwidawsk.net \
--cc=intel-gfx@lists.freedesktop.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