From: Laszlo Ersek <lersek@redhat.com>
To: "Denis V. Lunev" <den@openvz.org>, qemu-devel@nongnu.org
Cc: "Denis Plotnikov" <dplotnikov@virtuozzo.com>,
"Marc-André Lureau" <marcandre.lureau@gmail.com>,
"Michael Roth" <mdroth@linux.vnet.ibm.com>,
"Stefan Weil" <sw@weilnetz.de>
Subject: Re: [Qemu-devel] [PATCH v4 1/1] qga: minimal support for fstrim for Windows guests
Date: Wed, 5 Oct 2016 20:55:22 +0200 [thread overview]
Message-ID: <ed9df338-38c6-a4bc-8f69-992d6914e1f9@redhat.com> (raw)
In-Reply-To: <1475503285-9021-1-git-send-email-den@openvz.org>
On 10/03/16 16:01, Denis V. Lunev wrote:
> Unfortunately, there is no public Windows API to start trimming the
> filesystem. The only viable way here is to call 'defrag.exe /L' for
> each volume.
>
> This is working since Win8 and Win2k12.
>
> Signed-off-by: Denis V. Lunev <den@openvz.org>
> Signed-off-by: Denis Plotnikov <dplotnikov@virtuozzo.com>
> CC: Michael Roth <mdroth@linux.vnet.ibm.com>
> CC: Stefan Weil <sw@weilnetz.de>
> CC: Marc-André Lureau <marcandre.lureau@gmail.com>
> ---
> qga/commands-win32.c | 97 ++++++++++++++++++++++++++++++++++++++++++++++++++--
> 1 file changed, 94 insertions(+), 3 deletions(-)
Just to educate myself (really, you can ignore my question as "review
comment", because it's not one): why is this necessary? In Windows 8 and
Windows Server 2012 R2, simply putting your NTFS filesystem on a SCSI
disk (such as virtio-scsi-pci / scsi-hd) or AHCI, and specifying
discard=on for the -drive, will result in the guest automatically
trimming the NTFS (with a little delay) after deleting files, and the
host getting those blocks back.
Thanks
Laszlo
> Changes from v3:
> - fixed memory leak on error path for FindFirstVolumeW
> - replaced g_malloc0 with g_malloc for uc_path. g_malloc is better as we are
> allocating string, not an object
>
> Changes from v1, v2:
> - next attempt to fix error handling on error in FindFirstVolumeW
>
> diff --git a/qga/commands-win32.c b/qga/commands-win32.c
> index 9c9be12..cebf4cc 100644
> --- a/qga/commands-win32.c
> +++ b/qga/commands-win32.c
> @@ -840,8 +840,99 @@ static void guest_fsfreeze_cleanup(void)
> GuestFilesystemTrimResponse *
> qmp_guest_fstrim(bool has_minimum, int64_t minimum, Error **errp)
> {
> - error_setg(errp, QERR_UNSUPPORTED);
> - return NULL;
> + GuestFilesystemTrimResponse *resp;
> + HANDLE handle;
> + WCHAR guid[MAX_PATH] = L"";
> +
> + handle = FindFirstVolumeW(guid, ARRAYSIZE(guid));
> + if (handle == INVALID_HANDLE_VALUE) {
> + error_setg_win32(errp, GetLastError(), "failed to find any volume");
> + return NULL;
> + }
> +
> + resp = g_new0(GuestFilesystemTrimResponse, 1);
> +
> + do {
> + GuestFilesystemTrimResult *res;
> + GuestFilesystemTrimResultList *list;
> + PWCHAR uc_path;
> + DWORD char_count = 0;
> + char *path, *out;
> + GError *gerr = NULL;
> + gchar * argv[4];
> +
> + GetVolumePathNamesForVolumeNameW(guid, NULL, 0, &char_count);
> +
> + if (GetLastError() != ERROR_MORE_DATA) {
> + continue;
> + }
> + if (GetDriveTypeW(guid) != DRIVE_FIXED) {
> + continue;
> + }
> +
> + uc_path = g_malloc(sizeof(WCHAR) * char_count);
> + if (!GetVolumePathNamesForVolumeNameW(guid, uc_path, char_count,
> + &char_count) || !*uc_path) {
> + /* strange, but this condition could be faced even with size == 2 */
> + g_free(uc_path);
> + continue;
> + }
> +
> + res = g_new0(GuestFilesystemTrimResult, 1);
> +
> + path = g_utf16_to_utf8(uc_path, char_count, NULL, NULL, &gerr);
> +
> + g_free(uc_path);
> +
> + if (gerr != NULL && gerr->code) {
> + res->has_error = true;
> + res->error = g_strdup(gerr->message);
> + g_error_free(gerr);
> + break;
> + }
> +
> + res->path = path;
> +
> + list = g_new0(GuestFilesystemTrimResultList, 1);
> + list->value = res;
> + list->next = resp->paths;
> +
> + resp->paths = list;
> +
> + memset(argv, 0, sizeof(argv));
> + argv[0] = (gchar *)"defrag.exe";
> + argv[1] = (gchar *)"/L";
> + argv[2] = path;
> +
> + if (!g_spawn_sync(NULL, argv, NULL, G_SPAWN_SEARCH_PATH, NULL, NULL,
> + &out /* stdout */, NULL /* stdin */,
> + NULL, &gerr)) {
> + res->has_error = true;
> + res->error = g_strdup(gerr->message);
> + g_error_free(gerr);
> + } else {
> + /* defrag.exe is UGLY. Exit code is ALWAYS zero.
> + Error is reported in the output with something like
> + (x89000020) etc code in the stdout */
> +
> + int i;
> + gchar **lines = g_strsplit(out, "\r\n", 0);
> + g_free(out);
> +
> + for (i = 0; lines[i] != NULL; i++) {
> + if (g_strstr_len(lines[i], -1, "(0x") == NULL) {
> + continue;
> + }
> + res->has_error = true;
> + res->error = g_strdup(lines[i]);
> + break;
> + }
> + g_strfreev(lines);
> + }
> + } while (FindNextVolumeW(handle, guid, ARRAYSIZE(guid)));
> +
> + FindVolumeClose(handle);
> + return resp;
> }
>
> typedef enum {
> @@ -1416,7 +1507,7 @@ GList *ga_command_blacklist_init(GList *blacklist)
> "guest-get-memory-blocks", "guest-set-memory-blocks",
> "guest-get-memory-block-size",
> "guest-fsfreeze-freeze-list",
> - "guest-fstrim", NULL};
> + NULL};
> char **p = (char **)list_unsupported;
>
> while (*p) {
>
next prev parent reply other threads:[~2016-10-05 18:55 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-10-03 14:01 [Qemu-devel] [PATCH v4 1/1] qga: minimal support for fstrim for Windows guests Denis V. Lunev
2016-10-04 13:43 ` Marc-André Lureau
2016-10-05 11:13 ` Denis V. Lunev
2016-10-25 23:53 ` Michael Roth
2016-10-05 13:51 ` Richard W.M. Jones
2016-10-05 18:55 ` Laszlo Ersek [this message]
2016-10-05 19:47 ` Denis V. Lunev
2016-10-05 19:56 ` Laszlo Ersek
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=ed9df338-38c6-a4bc-8f69-992d6914e1f9@redhat.com \
--to=lersek@redhat.com \
--cc=den@openvz.org \
--cc=dplotnikov@virtuozzo.com \
--cc=marcandre.lureau@gmail.com \
--cc=mdroth@linux.vnet.ibm.com \
--cc=qemu-devel@nongnu.org \
--cc=sw@weilnetz.de \
/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 NNTP newsgroup(s).